Should we integrate SQLModel into the package?

[havok2063]

Should we integrate SQLModel into sdssdb for the SQLAlchemy models? I don't think this is strictly necessary for 1.0, but maybe this is something to consider for later. If we don't want a complete integration here, perhaps a smaller wrapper library could be considered.

Pros

• Automatically creates pydantic models for validation / serialization for all our ORMs.
• No need to duplicate code in downstream packages
• Makes it easier to drop into future python API services / frameworks

Cons

• Increases the dependency stack
• No integration with peewee models

[albireox]

I like the idea of exploring SQLModel. I haven't played a lot with it but it seems like a good idea and probably integrates well with things like FastAPI. I would leave this for after 1.0 but don't see any reason no t to include them for at least a subset of the schemas.

Maybe also interesting to discuss in the context of SQLA vs Peewee. I still like Peewee's simplicity, although haven't really looked too much into SQLA, but development seems to have stalled a bit and I think it's unlikely the developer will implement things like asyncio support (although it's possible that this could be implemented with very little change to sdssdb).