chore: Reenable SQLite tests which leverage foreign key constraints et al.

[john-bodley]

SUMMARY

In #24488 I added logic to enable foreign key support for SQLite. A positive byproduct of said change is there are a slew of SQLite integration tests which were disabled due to the lack of foreign key support (and/or blind copypasta).

This PR simply reenables the previously skipped SQLite tests.

BEFORE/AFTER SCREENSHOTS OR ANIMATED GIF

TESTING INSTRUCTIONS

CI.

ADDITIONAL INFORMATION

• Has associated issue:
• Required feature flags:
• Changes UI
• Includes DB Migration (follow approval process in SIP-59)
  • Migration is atomic, supports rollback & is backwards-compatible
  • Confirm DB migration upgrade and downgrade tested
  • Runtime estimates and downtime expectations provided
• Introduces new feature or API
• Removes existing feature or API

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (9033e72) 69.14% compared to head (4651688) 69.14%.
Report is 11 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master   #24605   +/-   ##
=======================================
  Coverage   69.14%   69.14%           
=======================================
  Files        1946     1946           
  Lines       75989    75989           
  Branches     8479     8479           
=======================================
  Hits        52544    52544           
  Misses      21266    21266           
  Partials     2179     2179           

Flag	Coverage Δ
hive	53.68% <ø> (ø)
mysql	78.05% <ø> (ø)
postgres	78.17% <ø> (ø)
presto	53.63% <ø> (ø)
python	82.86% <ø> (ø)
sqlite	77.76% <ø> (+0.93%)	⬆️
unit	55.80% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[john-bodley]

See here, specifically it mentions,

SQLite locks the database when a write is made to it, such as when an UPDATE, INSERT or DELETE is sent. When using the ORM, these get sent on flush. The database will remain locked until there is a COMMIT or ROLLBACK.

hence the need for the commit to free the lock prior to dropping the actual table.

[john-bodley]

Weird name for a database used for integration (as opposed to unit) tests. Calling this superset.db is inline with how the metadata database is named elsewhere—including in this workflow for MySQL and PostgreSQL.

[michael-s-molina]

Not a blocker for this PR but there's a task in 4.0 to drop support for SQLite to materialize SIP-33.

[john-bodley]

This threw me for a curveball. I was able to successfully run the tests locally but they failed remotely with a somewhat cryptic error stating a reference table being specified in this fixture didn't exist.

I speculate that this is probably due to a multithreading or connection pooling issue when running via a GitHub workflow. Ensuring a single connection per thread seems to remedy the problem. 🤞

[john-bodley]

Per here, if undefined the metadata and example databases are the same and given that SQLite doesn't support schemas. It seems prudent to split these into two after @dpgaspar's #25003 which isolated the examples database.

[craig-rueda]

LGTM.