Async SQLAlchemy support

[ermaruskin]

Is your feature request related to a problem? Please describe.
Currently, clickhouse-connect does not support using SQLAlchemy with async operations. Users must choose between get_async_client() for async capabilities or SQLAlchemy for query building, but cannot use both together. This is problematic for analytics applications that need SQLAlchemy's dynamic query construction (especially for filter-based queries) while maintaining async performance benefits.

Describe the solution you'd like
I would like clickhouse-connect to provide async SQLAlchemy support by implementing an async dialect that works with SQLAlchemy's create_async_engine().

Ideally, this would work similar to how other async database drivers operate:

engine = create_async_engine('clickhousedb+async://user:password@host:port/database')

Describe alternatives you've considered
The only alternative for asynchronous requests I could find is clickhouse-sqlalchemy which is unfortunately no longer maintained. Under the hood, another outdated library asynch is used.

[joe-clickhouse]

Hi @ermaruskin! You're right, unfortunately this is a feature gap today. Our driver has been around a while now and was originally designed sync-first, and we've since built an async wrapper around it (using ThreadPoolExecutor) that allows it to be used in async runtimes. We could implement something similar for our SQLAlchemy dialect which, of course, wouldn't benefit from being async-native but would at least bridge the gap and allow it to be used in async runtimes in the short term. A true async-first driver and SQA dialect is something we definitely need though and is on our radar and being prioritized internally. Thanks for raising this!