ADBC support?

[pimvh]

Hi! Thanks for creating this project!

I am wondering, have you considered supporting ADBC as a connecting interface towards the different databases? ADBC is relatively new, but read for production imo. I think introducing ADBC would:

1. simplify the codebase, and ease the maintenance of the several engine adapters, as they share the similar dbapi interface and have less non-shared dependencies.
2. improve the preformance of fetchdf when moving large amounts of data.
3. allow a BYOD (bring your own driver) approach.

Some pointers:
https://arrow.apache.org/adbc/current/python/recipe/postgresql.html#using-pandas-and-adbc
https://docs.adbc-drivers.org/drivers/bigquery/index.html

The difference between for example BigQuery, Databricks and DuckDB Quack and Trino then becomes:

from adbc_driver_manager import dbapi

dbapi.connect(
    driver="bigquery" # driver="databricks" or driver="quack" or driver="trino"
    db_kwargs={
       # these can be passable via the config and vary per database and user requirements
       # "uri": "quack://localhost:9494/?token=quack-secret"
    },
)

I think it would simplify managing models in multiple databases/datalakes, and enhances/leverage the existing power of SQLGlot even more.

I'd have to dive deeper into the code to figure out the overall impact on the codebase though.