Rework to allow use of metadata

[MatBarba]

Maybe too many things changed, but I needed those to ensure I can use a metadata with :

• DBConnection can now be created without a schema (dump_dir no longer mandatory)
• DBConnection accepts an Metadata object instead of relying on an SQL file and a reflection
• Add 2 methods to DBConnection: create_all_tables and create_table (requires the metadata to be set as above)
• If both a dump_dir and a metadata are provided to a UnitTestDB, the schema is made from the metadata and the data is loaded from the dump_dir
• Add a context manager to properly dispose of the UnitTestDB (UnitTestDBContext)

Tests:

• Add a simple mock base and mock_metadata
• When possible, use a context session_scope or test_session_scope to avoid hanging connections (unless we are specifically testing for that)
• Both SQLalchemy and MySQL should work
• Add an argument tmp_path to the UnitTestDB when asked to create an SQLite database test file

Usage:
Say you want to test interactions with a whole schema described in a given metadata. You can do so with:

with UnitTestDBContext(server_url, metadata=metadata, tmp_path=tmp_path, name="test_metadata") as test_db:
    with test_db.dbc.test_session_scope() as session:
        results = session.execute(text("SELECT * FROM foo"))

The first context ensures the database is properly disposed of no matter what, and the second ensures the changes are rolledback and the connection returned.

[JAlvarezJarreta]

Quite a few interesting changes indeed. I quite like the approach and functionality proposed, great addition 👍

I have added a few suggestions, changes and questions for discussion. I believe the 2 new test methods in test_unittestdb.py could be merged into one, as it feels the functionality and objectives are very similar.

[JAlvarezJarreta]

We got there at the end!