shared base implementation for ZetaSQL to be used by Spanner, BigQuery, Apache Beam SQL?

[tswast]

Discussing the possibility of using this package from Ibis (ibis-project/ibis-bigquery#72), I realized it may be helpful to share code between BigQuery and Spanner (and other Zeta SQL dialects) in the SQLAlchemy connectors, similar to what we've done (but haven't completed) with the Ibis connectors.

CC @larkee -- I recall you mentioned there was work being done on a SQLAlchemy connector for Spanner? Perhaps we can collaborate on a shared ZetaSQL SQLAlchemy base implementation?

[tswast]

https://github.com/cloudspannerecosystem/python-spanner-sqlalchemy -> https://github.com/googleapis/python-spanner-sqlalchemy?

I'm not sure if it's even possible to make something useful in SQLAlchemy without connecting to a real data system, but it's worth investigating.

[tswast]

I learned a bit more about the Spanner connector since I filed this issue. They made some choices in their DB-API where the types of the values returned are different from what we do in BigQuery (and from what is expected by pandas). It'd be a breaking change for them to change at this point, so I'll close as infeasible.