Refactor calculation of annualized_respondents_ferc714 (#3024)

* Refactor calculation of annualized_respondents_ferc714

Rather than loading huge demand_hourly_pa_ferc714 dataset and calculating
report_date columns from these, we can infer these values from the ferc714_settings.
Additionally, we can use cross product merge to blow out the respondents, rather
than doing the complex procedure that we did up to this point.

* Put back check that report_date is not in respondent_id_ferc714
* import dataset_settings resource into annualized_respondents_ferc714
* Update conda-lock.yml and rendered conda environment files.

---------

Co-authored-by: rousik <[email protected]>
Co-authored-by: Zane Selvans <[email protected]>
Co-authored-by: zaneselvans <[email protected]>