feat(orm): validate connection config at DatabaseManager construction time

[tmgbedu]

Summary

Follow-up to backlog task #1223 / PR #194, per the reviewer's recommendation on that PR (task #1231).

PR #194 hardened ConnectionFactory.build_url()/make() so a bad connection config fails with a friendly DriverNotFound instead of a raw KeyError/ValueError -- but only at the point of first connection/engine-creation (e.g. db:migrate, or the app's first query), since that's when factory code first runs.

This PR moves that failure earlier: DatabaseManager now validates every configured connection up front, in __init__. Since DatabaseProvider.register() constructs the DatabaseManager during app boot, a misconfigured connection now raises immediately when the Application is constructed, not lazily on first DB use.

Validated per connection (skipped entirely if the connection supplies an explicit url, mirroring build_url()'s own passthrough):

• driver key must be present -> DriverNotFound if missing
• driver value must be one of ConnectionFactory.DRIVER_URLS -> DriverNotFound if unsupported (reuses the exception + supported-driver list from PR fix(orm): harden ConnectionFactory against bad port and unknown driver #194, doesn't reinvent)
• for non-sqlite drivers, a port that's present but not int-castable (e.g. "not-a-port") -> ValueError. A missing/empty port is not an error since ConnectionFactory already defaults it per-driver.

Test plan

• uv run pytest fastapi_startkit/tests/masoniteorm/config/test_manager.py -v -- new TestDatabaseManagerBootTimeValidation class: happy path (sqlite/mysql/postgres all construct fine), missing driver key, unsupported driver, non-numeric port, missing port is not an error, explicit url bypass, and an end-to-end Application boot repro -- all passing
• Updated test_connection_raises_for_missing_driver in test_model_attributes.py to assert the failure now happens at DatabaseManager(...) construction rather than at first .connection() call
• Full suite: uv run pytest --ignore=tests/masoniteorm/postgres from fastapi_startkit/ -- 1988 passed, 7 skipped, no regressions
• Coverage: 81.25% (threshold 80%)
• Manually verified: a valid config still runs db:migrate end-to-end unchanged; a bad config (unsupported driver) now raises DriverNotFound at Application(...) construction, before db:migrate or any query ever runs
• ruff check / ruff format --check clean

Per the current merge freeze, this PR is intentionally left open/unmerged.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!