They were getting reset to their latest reply date, when a bumplocked thread should never have its
bump date reset to earlier than whatever the latest reply was when it was bumplocked, or the OP date
if all replies are deleted.
Still not accounting for bump limited thread (a uncommon edge case), and i think the simplest way to deal with that would
be set bumplocked:1 when its exceeded the board settings bump limit. More efficient, less work for me, and actually
serves a useful purpose because long threads could tell when they are bumplocked from going past bump limit.
The only decision then would be whether to set it permanently from that point forward, or only set it once,
so they could be unlocked by moderators if they want which would be an added side effect/feature of being able
to selectively unlock them.
wtf is this commit message, and the code for this section is a cluster fuck lmao
Previously early404 would match the threads with less than "early404Replies" BEFORE skipping and then deleting.
Which would be incorrect because then early404 would only be deleting a fraction of the posts with less than
early404Replies, rather than any thread with less than early404Replies past a certain fraction of the threadLimit
Bit of a stupid mistake, but this should fix it, if my interpretation of how early404 is supposed to work is correct.
This prevents prolems like `/` giving 404 in devel mode (when
`static/html/index.html` is missing) or `/captcha` redirecting to
`/captcha/` (then breaking).
* consolidate so only a single aggregate is needed with better grouping in deleting posts to fix reply counts and bump dates are also not fixed on posts deleting
* changes to account for empty threads and bulkwrite results
* some changes, but date changing rebuilds could build redundant pages, so gona ditch this for now
* make thread bounds expand for when deletes that change date happen so shadow bumps dont occur and correct pages get rebuilt
* add board to empty aggreagte for resetting reply counts to use to prevent shadow bumps