[FEAT] Initializes daft-sql and defines the daft.sql(..) function.

[RCHowell]

This pull request adds the daft-sql module to the Rust side, and defines a daft.sql(..) method along with a placeholder test.

Then intention of this PR is to establish the structure and plumbing before introducing any functionality.

[RCHowell]

Demo

import daft
from daft.sql.sql import SQLCatalog

# register dataframes to the catalog
catalog = SQLCatalog({
  "example": daft.from_pydict({ "a": [1, 2, 3] }),

})

# invoke daft.sql()
df = daft.sql("SELECT * FROM example", catalog)
df.collect()

[universalmind303]

so i think the work that I did in #2558 will compliment this pretty well.

#2558 adds a lot of sql parsing logic, and this PR sets up a lot of the python frontend that #2558 is missing.

[RCHowell]

Nice! I saw that we started to overlap on the Rust side — you were able to get a lot done so quickly. If you want to focus on the Rust side, I will rebase from #2558 when I have bandwidth. Otherwise, you are welcome to take the Python frontend things you wish and change up the names like SQLCatalog->SQLContext.

[universalmind303]

Nice! I saw that we started to overlap on the Rust side — you were able to get a lot done so quickly. If you want to focus on the Rust side, I will rebase from #2558 when I have bandwidth.

@RCHowell no need, we can merge this in first and I can rebase off of yours.

[universalmind303]

nit: is there any way we can have this just from daft.sql instead of `from daft.sql.sql

[RCHowell]

I totally agree, but struggled without moving existing things. I had a circular import error by adding things in sql.py to the sql/__init__.py because of the existing files.

Is moving sql_scan and sql_connection to avoid the circular dependency a viable option?

[universalmind303]

a few small nits, but overall looks good! great work @RCHowell!