test flakiness: some tests just get stuck

[callebtc]

we don't know why and it's making us crazy

https://github.com/cashubtc/nutshell/actions/runs/23343685006/job/67903379596?pr=874

more flakiness
https://github.com/cashubtc/nutshell/actions/runs/23642408319/job/68866093664

[b-l-u-e]

we don't know why and it's making us crazy

https://github.com/cashubtc/nutshell/actions/runs/23343685006/job/67903379596?pr=874

the fact that in tests/conftest.py

        db = Database("mint", settings.mint_database)
        async with db.connect() as conn:
            # drop all tables
            await conn.execute("DROP SCHEMA public CASCADE;")
            await conn.execute("CREATE SCHEMA public;")
        await db.engine.dispose()

each of 16 tests drops or recreates the entire postgres schema, running migrations, initializes the corelightningrest backend and starts up the ledger

regtest-mint hit the 10-minute job timeout and was cancelled while pytest tests/mint was still running

1. Is 120 minutes acceptable for now or should we measure the slowest matrix cell and set a tighter value?
2. In long term, should we reduce work in tests/conftest.py ledger something cheaper than DROP SCHEMA per test where tests allow it?
   3.Should a lighter regtest matrix run on every PR, with the full matrix on main or nightly?