No longer fail on the first error, but collect them all.

[McModknower]

This is currently a draft, since i am not that happy with the implementation yet.

TODO:

• Do this for the other modloaders as well.
  Maybe via a bit more generic way.
• Should some errors (DNS, full disk, rate limit exceeded) still cause direct stopping?
• Test the additions (maybe via python -m http.server on localhost by changing the domain for testing?)

References https://discord.com/channels/132965178051526656/914690317033300018/1441134150571659284

[peterix]

That looks quite reasonable.

Should some errors (DNS, full disk, rate limit exceeded) still cause direct stopping?

DNS => most likely needs to fail
full disk => ideally doesn't happen, but if it does, what is this thing even going to do? It's not built to handle that. Maybe check free space at start and if there is less than N GiB, fail fast?
rate limit => could be handled by backoff retry scheme of some sort, but it's not in scope for this simple implementation IMO.

Like, it's been reliable enough as it is for a very long time. If we were tracking a lot more projects, the probability of something returning an error would increase, and then dividing the meta update up into smaller parts and handling more errors better would definitely make sense.

But yeah, I'll always be in favor of collecting all errors and presenting them in a list. There's nothing more annoying than having to rerun a long running script just to see the next problem (ad nauseam).