This is a suggestion for further research, so it will by necessity be a stub of an answer. My plan would be to set the records aside for a period of time, and then review all the collected material as a group -- and here's the important part -- with the assumption that each record belongs to a different George Wills until you have information that suggests otherwise.
Services like Ancestry and Findmypast encourage us to match records by name, and there is a tendency to match up records and presume they belong to "our people" in advance of doing a thorough analysis of what we have.
Take each record and examine it using the guidance in Elizabeth Shown Mills' Evidence Analysis Process Map. Keep in mind her advice in QuickLesson 1: Analysis & Citation.
Each time we find a source, we must analyze it to determine what it is. Then we identify it. That physical analysis and identification of our source is expressed as a citation. At publication stage, that citation is typically stripped down to the bare minimum needed to relocate the source. But, in the research phase, the details we capture in that citation provide the foundation for evaluating or analyzing the information that source provides.
For instance: when looking at the Bodmin Gaol Records at Ancestry, there is no image -- the graphic says "No Image / Text-only collection". Ancestry also gives us this note:
Original data: Data provided by Patricia T. Fawcett and Sally J. Pocock. Cornish Prison Records. Cornwall Record Office, Truro, Cornwall, England.
You say "I have only seen the transcript...." But we don't have the original images to compare this data to, so we can't say if this a transcription -- it might be more accurate to say this is an extract from the original records. In particular, we don't know what the asterisk means, and I don't know how we could possibly tell from looking at only this database unless we looked at the entire database and correlated the entries with other records, or compared the extracts to the images.
So whenever I find records on any website, the next thing I do is try to find information from the archivists who actually hold the records, to see if I can learn more. All three of the extracts you cited are from Cornwell Record Offices record group AD 1676, Cornwall Gaol Records -- link to their catalog page.
At the end of the scope and content listing, it says:
Many prisoners were transported, for terms of seven to twenty years and the register gives details of the hulks to which they were sent to await transportation. [Information from (and see also) 'The Inmates of Cornwall County Gaol at Bodmin' by Pocock, S, in the Cornwall Family History Society Newsletter number 123, March 2007.]
The catalogue entry tells us that when people were admitted to the Bridewell multiple times, sometimes the later entries will refer back to the first entry number. You can see these notations in the entries for George Wills from Truro that turn up when searching by name at Ancestry.
Here are all the search results for an exact name search, including the ones you've already discussed:

Let's look at George Wills from Truro to compare and contrast the entries for your George from Illogan late Redruth. Putting them in order of admission, we see:
- 1866: Reg Number 416, Notes: Removing straw from parish of St. Erme to parish of St. Clement contrary to Qtr. Sessions Order made Feb. 1866. Scar on right elbow. R.C.R. M[ilitia]
- Dec 1870: Reg No. 5545, Notes: M+1 2nd time, see 418 RCRM
- Jul 1871: Reg No. 5603, Notes: M 3rd time, see 5545
These three entries clearly belong to the same person, since the later entries are cross-referenced to the earlier offenses. Note that in all three, there are new registration numbers each time (as opposed to what you can see in the National School Admissions registers, where students are re-admitted with the same number instead of being assigned a new number in the sequence).
Now let's look at the remaining search results. The two search results for George Wills, Reg No 122, seem to refer to the same man -- one is for a man from Devon, and the other has this note:
See separate list of foreign seamen and men from other counties and countries
For the remaining records, I would agree with the previous answer that the records for George Wills, registration number 4290, who was convicted and sentenced for 7 years for Stealing Pork all refer to the same person. We see from the extract that one of your George Wills was sent to the Hulk H.M.S. Captivity. Of the Captivity, Wikipedia says:
The ship was taken into Sheerness Dockyard in April 1826, and was
fitted out for the journey to Plymouth. She arrived there in June
and spent the last eight years of her working existence as a convict
hulk in Plymouth. By 1834 the rate of penal transportation had been
drastically increased to clear out the old hulks. When the last
convicts had left Captivity, she was handed back to the Navy
Department, who put her up for sale.
The Commissioners for executing the office of Lord High Admiral of the
United Kingdom of Great Britain and Ireland gave notice that they
would offer several vessels for sale on 21 January 1836, including
Captivity, of 1613 tons, then lying at Plymouth. She sold on that
day for £4,030. Advertisements in the local Plymouth,
Devonport and Stonehouse News in September 1836 announced the
auctioning off of her timber.
If her timber was auctioned off in 1836, what happened to the prisoners whose terms hadn't yet expired? This is earlier than the discharge date you found of 18 Dec 1838.
It might be worthwhile to take a look at the registers again and compare the entries for George Wills to the other prisoners in the same registers and to the other prisoners named George Wills who are a different age or committed a different offense, just to get a feeling for the records. If you have 26 quarterly entries in the registers, that does seem consistent with an intake date in July 1832 and a December 1838 discharge.
That leaves George Wills, Reg No. 5941. A register entry for an offense in 1839 is consistent with someone being released in December 1838, if it is a new offense -- but we have no cross-reference in our online database. If we could look at the original entries, we might get a physical description, but since we aren't in the Cornwall Record Office, how could we get more information, either to help us decide whether to order the records, or while we are waiting for record copies to arrive?
Search for other source types
Another issue with the question is that you've started from the standard base of church records, civil records, and census records. Now you're trying to evaluate the Gaol records against this bare framework. For criminal records, the easiest way into finding the information that may have been discarded or is otherwise lost to us -- what happened during the trial -- is to look for newspaper accounts of what happened during the criminal sessions. Note that in one of your entries, you have this crime:
Assaulting Jn. Teague, mine agent. Acq.
An assault on a mine agent sounds like just the ticket for a newspaper editor to fill the 'newshole' on a slow news day, and you have the advantage of another name to look for in addition to the accused.
Put all of the records you have so far into a timeline, sorted into different columns where you can determine you have a group of records that belong with each other. Then mine your records for clues, starting with the records at the time of death (or later, if you have probate records) and working backwards. Investigate any details you find and gather more information for context as needed.
Some records at TNA (referred to as Digital Microfilm) are free to view online and to download. Some of these records from the Home Office (HO) deal with prisons and convict records, including the hulks. A search for "hulk + Captivity" yielded 154 results including petitions for mercy for lists of prisoners and other correspondence. Some of the catalog entries list the names of individual prisoners and others merely mention the name of the hulk. Filtering those results by 'Available for download only' plus the time period of 1800-1899 reduced this list to two record groups:
- HO 9/3: Description: HO 9. Convict hulks moored at Woolwich and Devonport: Captivity, Ganymede, Discovery: Register of
prisoners. Date: 1821-1833
- HO 9/8: Description: HO 9. Convict hulks moored at Portsmouth: Portland, Captivity, Leviathan: Register of prisoners.
Date: 1802-1836
There may be other freely-available records in other series, and other records at TNA regarding this case, but these two free collections seem to be right in the ballpark for this question.
UPDATE: see TNA's research guide on Criminals courts and prisons and the search page at findmypast for England & Wales, Crime, Prisons & Punishment, 1770-1935 for links to further resources. In the right-hand column on the page, one of the links under Useful links and resources is Prison Ship (Hulk) Registers which will take you to the search form to search registers dated between 1811 and 1843 which have been indexed by findmypast.