feat(cache): Infrastructure to distinguish between errors caused by malformed objects vs cache-specific errors

[relaxolotl]

This is part 3 of a series of PRs focused on cache statuses. Like the others, if you haven't taken a look at the previous PRs and have no context on these changes, it may be helpful to take a look at those first: #509 #510

1. Introduce the new cache status without changing the existing functionality, convert the new status to Malformed if needed (feat(cache): Add a new cache status CacheSpecificError #509)
2. Update cache expiration strategy selection so different caches will expire entries triggered by cache-specific problems after a timeout that makes sense for the cache (CacheStatus::CacheSpecificError entries have different timeout behaviours based on the type of cache they belong to #516)
3. Extend both the new cache status and the Malformed cache status to accept an arbitrary string that describes the root cause of the issue (feat(cache): Stuff error messages into Malformed and CacheSpecificError cache files #510)
4. 👉 Identify and split out code paths that should be using the new cache status without changing their return value (feat(cache): Infrastructure to distinguish between errors caused by malformed objects vs cache-specific errors #511)👈 you are here
5. Begin actually using the new cache status by returning it in places identified in the previous step and writing + reading those to and from the cache (feat(cache): Use the CacheSpecificError cache status #512)

This PR is focused on the downloaders. It adds documentation and extends them as needed so it's easy to (eventually) tell the difference between a malformed object and a cache-specific error. This PR makes no functional changes to symbolicator, and simply makes some changes in preparation for step 4.

#skip-changelog

[Swatinem]

no need to do that in this PR, but the debug logs here print all of the source information, including private S3 keys, etc to the console, so that is waaay to detailed.

[relaxolotl]

i'll create a follow-up PR that tries to address this in a simple way 👍