Adding cache schema version

[EngCaioFonseca]

Pull Request Change Description

This solves #1198 and adds the cache schema versioning as well as lay the groundwork for the follow up steps or making the cache persistent.

The cache schema is now versioned with Alembic. It runs on startup: a fresh cache gets stamped at the latest version, and an existing cache gets any missing migrations applied automatically. Base tables are still created with the raw CREATE blocks in db_init.py, so adding a new table works exactly like before. Alembic only handles changes to existing tables.

The first migration adds the labels column from #1189. Existing caches never got it, because CREATE TABLE IF NOT EXISTS doesn't touch a table that already exists. The migration adds the column and clears the old cached issues so they get re-collected with labels.

A few things added for the cache long-term:

• build the repo_id indexes without locking the tables (CREATE INDEX CONCURRENTLY)
• only flush the Redis broker when the cache actually changed instead of on every startup, and stopped it from also wiping user sessions
• take a lock during init so two app pods starting at once don't step on each other
• build the DB connection strings with psycopg2's make_dsn so passwords with special characters don't break them

This is also groundwork for making the cache persistent instead of rebuilt on every boot.

Generative AI disclosure

Please select one option:

• This contribution was NOT assisted or created by Generative AI tools.
• This contribution was assisted or created by Generative AI tools.

If AI tools were used, please provide details below:
- What tools were used? Claude - Opus 4.8; GPT - 5.6 Sol
- How were these tools used? Draft, review iteration.
- Did you review these outputs before submitting this PR? Yes.

[MoralCode]

#1189 made changes to the schema, so this may need to change to account for that

[MoralCode]

can you elaborate on why we aren't using alembic? This seems like a very odd technical decision to me, especially if we are using SQL functions in its place

[EngCaioFonseca]

Fair point, and I kinda agree, but in this we're not removing alembic, or touching it in this repo. I've updated the comment to explain the reasoning behind it. In summary, since it is a small fixed schema, as well as disposable/unlogged and rebuildable cache, it's easier and more efficient to not use it and it's the reason why we don't need it.