add SQL adapter

[skarakuzu]

preliminary start of sql adapter. to be continued ...

Checklist

• Add a Changelog entry
• Add the ticket number which this PR closes to the comment section

[danielballan]

This section is used when tiled is installed like pip install "tiled[all]". These three should also be added to the section server, below, so that they are included when tiled is installed like pip install "tiled[server]" # server only.

[danielballan]

Lifecycle:

1. Client declares that it wants to create a new tabular dataset, via a request POST /api/v1/metadata/my_table.
2. In "catalog" SQL database, the server adds a row to the nodes table with any metadata about this table. This is how the new table is connected to any overall dataset, like Bluesky scan and its Scan ID.
3. Also in the "catalog" SQL database, the server adds a row each to the data_sources table and the assets table. Together, they describe how to locate where the new data will be saved. The Asset part is very locked down. It has room for the URI of the tabular SQL database: postgresql://... and some boilerplate. The DataSource has a freeform area called parameters, which can fit any JSON. We can use this to put in dataset-specific details, like the name of the SQL table (table_name)---derived from the Arrow schema in this case---and a means of selecting the rows of interest for this new dataset (dataset_id).
4. When data is written or read, a SQLAdapter object is instantiated inside the server. It is passed information extracted from this DataSource and Asset. So, it can know the table_name and the dataset_id.