Python cql2 to_sql uses generic dialect, breaking A_OVERLAPS for Postgres

The Python bindings only exposes `to_sql` which transforms to SQLs generic dialect. Uses with Postgres, this works almost, but at least A_OVERLAPS which translates to `@@` isn't correct (should be `&&` in Postgres).

I believe this isn't meant to support all dialects, probably something like exposing `to_sql_ast` would be the right way so users can build on top of that to derive their correct dialect SQL?

Start by tracing the Python binding's to_sql entry point and the handling of A_OVERLAPS. Investigate whether exposing to_sql_ast gives callers enough information to derive PostgreSQL-compatible SQL, then verify that the generic output no longer prevents correct PostgreSQL use. Done means the Python API supports the required dialect-specific path without breaking existing generic SQL behavior.

Written by the indexing model from the issue text.