Defer validation of PudlTabl datastore to eia861/ferc714 ETL methods

[zaneselvans]

@gschivley's users were having issues with PUDL being installed as a dependency, and relying on a raw Datastore being present, even though they're just working with data in the SQLite database, becuase of the hack we have in place now to provide direct access to the ETL outputs for ferc714 and eia861 inside the PudlTabl output object. This PR addresses that problem:

• Documented all of the inputs to the PudlTabl class, including the Datastore.
• Moved the instantiation of a default Datastore into the methods where it's actually required (etl_ferc714 and etl_eia861) within the PudlTabl class, so that anyone who is relying solely on a free-floating PUDL SQLite DB doesn't have to worry about having a working datastore PUDL workspace... until they want to access the data that actually depends on it.

[codecov]

Codecov Report

Merging #1275 (bae3905) into dev (11d95c4) will decrease coverage by 0.28%.
The diff coverage is 36.36%.

@@            Coverage Diff             @@
##              dev    #1275      +/-   ##
==========================================
- Coverage   81.25%   80.96%   -0.28%     
==========================================
  Files          54       54              
  Lines        6393     6409      +16     
==========================================
- Hits         5194     5189       -5     
- Misses       1199     1220      +21     

Impacted Files	Coverage Δ
src/pudl/output/pudltabl.py	74.81% <36.36%> (-2.74%)	⬇️
src/pudl/analysis/timeseries_cleaning.py	85.78% <0.00%> (-3.06%)	⬇️
src/pudl/workspace/datastore.py	70.45% <0.00%> (+1.62%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 11d95c4...bae3905. Read the comment docs.

[bendnorman]

Looks good to me! Just so I understand, this datastore logic in etl_eia861 and etl_ferc714 will be removed once they are included in the main etl?

[zaneselvans]

Yes that's right. This is just a hack because we weren't able to work with the EIA 861 under the current harvesting & entity resolution problem, but we got them to the point of being extracted and transformed and wanted the data to be available in some form.

Ultimately I think we want to be able to fully separate data production from data use, potentially even splitting the ETL package from a slimmed down user facing analytical package. Or a user might even just have the SQLite and Parquet files, and be working with some other toolset entirely.