Auto-Tagger might be sped up by keeping tags in RAM

[paleodictyon]

Describe the bug
The Auto-Tagger may be improved by holding tag info in memory rather than repeated individual SQL queries.
I was watching the logs while auto-tagging and saw tons and tons of SELECT alias from tag_aliases WHERE tag_id = ?, args: [1234]
Seems like when working with large data sets it should be checked to see if this is a bottleneck.
I haven't dug into the code, but there may be ways to improve the efficiency by writing some more of the logic in SQL, it may be worth exploring.

Stash Version: (from Settings -> About):
v0.20.2
Build hash:
e685f80
Build time:
2023-04-07 22:47:42