Feat/bigquery tracking

[claramoreirag]

LLMstudio Tracker Version 1.0.3

What was done in this PR:

• Added Tracker integration with BigQuery, using the sqlalchemy-bigquery to extend sqlalchemy support to BigQuery
• Bigquery does not support some SQL features, so the Table model needed to be adapted for compatibility:
  - autoincrement on IDs is not supported, so log id is given a value based on current_timestamp
  - JSON columns are not supported, so a custom type JSONEncodedDict was created on db_utils.py to encode JSON objects in strings before writing and decoding on read.
  - timestamp defaults were changed from server_default = func.now() to being set directly in python

How it was tested:

• Following the example in examples/04_bigquery_integration.ipynb

Additional notes:

Two dependencies were added to the pyproject.toml:

• sqlalchemy-bigquery
• google-cloud-bigquery-storage

[diogoncalves]

let's test the datetime functions