Update CsvExtractor to handle new 176/191/757A data format

[e-belfer]

EIA 176 data was originally getting pulled from the bundled data, after investigation of the bulk API data proved it was not suitable. In #2603 (comment) we discussed updating the data source to query NGQV directly. This has now been implemented in catalyst-cooperative/pudl-archiver#267, producing 3 new archives with 1 zipped CSV per year (176, 191, 757A archives). The advantage of this format is that we:

• don't have to hand-map line numbers to variables
• can easily extract a subset of years, as we do with other EIA data

However, the CsvExtractor written by @davidmudrauskas will have to be slightly updated to handle this new data format. The extractor should be able to take a subset of years from the ETL settings, pass it through dataset_settings to the extractor, and extract and concatenate the relevant zipped CSVs into one dagster asset. There should be three extractors with essentially identical implementations of the archiver - 176, 191 and 757A.