fix: Use singleton for database engine object for better connection pool

[egekocabas]

Motivation

We were getting the database connection pool error while we were bursting one endpoint that uses the database connection.

I searched about this error, and it seems that creating an engine for every time (every time get_engine() or get_session() is called) consumes all the connection pools.

Using a singleton pattern for the engine object and letting it manage its connection pool fixed the problem locally. I managed to send requests (uses the get_session()) to the /feed-raw-data endpoint with 5000 requests in a loop without a connection pool problem.

Changes

• get_engine() updated to use a single engine object. Pre-commit warning is suppressed because if we implement a class and then use its variable then we need to change every place where we use get_engine() and get_session(). I suggest to use it like that and then if it works and we have a time we can change the structure in here.

Checklist

• [x] added myself as assignee
• [x] correct reviewers
• [x] descriptive PR title using conventional commits.
• [x] description explains the motivation and details of the changes
• [x] tests cover my changes
• [x] my functions are fully typed
• [x] documentation is updated
• [x] CI is green
• [x] breaking changes are discussed with the team and documented in the PR title ! (e.g. feat!: Update endpoint)

[github-actions[bot]]

☂️ Python Coverage

current status: ✅

Overall Coverage

Lines	Covered	Coverage	Threshold	Status
2046	1500	73%	0%	🟢

New Files

No new covered files...

Modified Files

File	Coverage	Status
parma_analytics/db/prod/engine.py	100%	🟢
TOTAL	100%	🟢

updated for commit: 4f44750 by action🐍