Fix S3 adapter retrying failing uploads with exponential backoff

[Gargron]

The default limit of 10 retries with exponential backoff meant
that if the S3 server was timing out, you would be stuck with it
for much, much longer than the 5 second read timeout we expect.

The uploading happens within a database transaction, which means
a failing S3 server could negatively affect database performance

[smiba]

I know this is an old change, but could this in any way be the reason some people (me included) experience media sometimes being not available (as in #13739 #16520)?

Running tootctl media refresh on the posts fixes it, suggesting it's a temporary failure. There doesn't seem to be any attempt from mastodon to retry these posts though, they will stay at "Not Available" forever (until I run said command).

I'll have to dig a bit deeper in how mastodon handles it's media_cache in s3, but it's a recurring issue for me that doesn't seem to be logged either.

[smiba]

I want to confirm that setting the retry_limit to 2 has severely reduced my image failures to basically near 0%.
I've yet to see any again myself, tootctl media refresh barely does anything nowadays :)

What to do with this? Should we slightly increase the retry_limit until we have developed a proper solution for S3 upload failures? @Gargron