closing connection once data diff is executed

[sar009]

This PR revamps how we close database connections, adds new unit tests to the main entrypoint logic, and refactors tests for better readability.

[sar009]

fixes issue #784

[sar009]

I noticed you only added the close() method to postgres. You'll need to define that for the other databases. Should be straightforward for other files. Example: databricks.py, redshift.py, etc.

All other classes are handled from the close method in ThreadedDatabase or from Database since postgres.py was overriding close and has some special way of handling connection

[sar009]

Let's also add a test to test_database.py and assert that the close method is working as expected. It'll dynamically test all the databases are closing as expected.

I am not sure how this will dynamically test for all of the databases. For now, I have written for Postgres (https://github.com/datafold/data-diff/pull/785/files#diff-82fbc96aa3f082d4667c09a2a35aec050c73120de9174e1f37bef26ef9cd3115R173)

[sar009]

@sungchun12 I had to remove connection sharing since it was causing test failure in parallel unit test run
https://github.com/datafold/data-diff/pull/785/files#diff-53eb316ef7b18ba47acdd955cdaa7ca5467cd5ba6f5cf8ed92c68a421902d120L90

[sungchun12]

@sar009 any updates?