fix: add failback for shef infile import when tsid doesn't exist

[msweier]

cwmscli/commands/shef/import_infile.py

• Split store_group into _save (one store→update attempt) and store_group (orchestrator).
• Added _filter_existing_tsids, which probes each TSID with cwms.get_timeseries_identifier and drops missing ones with a warning.
• New flow: try the bulk save first; on failure, validate per-TSID, rebuild the payload from survivors, and retry. If every TSID validates, the original error re-raises (failure
  wasn't TSID-related).
• store_group now accepts entries and category_id so the retry can rebuild the JSON.

cwmscli/commands/commands_cwms.py

• Expanded the shef import_infile click help= so the auto-generated CLI reference describes the alias-id encoding and the new validate-and-retry fallback.

tests/commands/test_shef_import_infile_dry_run.py

• Added test_missing_tsid_falls_back_to_validated_save: 3-entry fixture, mocks store_timeseries_groups to fail when the missing TSID is in the payload and update_timeseries_groups
  to fail outright, mocks get_timeseries_identifier to raise for one TSID. Asserts two store calls (3 then 2 members), the missing TSID is absent from the retry, and the expected
  warnings are logged.

[msweier]

hang on....getting more in files to make this importer more robust

[msweier]

@Enovotny this is ready for review/merge. I forgot to follow up. Some of the issues we were having were group issues with empty timeseries. Should be good to go now

[Enovotny]

@msweier did you see a new version of shef-parsers that went out yesterday that should fix the empty ts issue?

[msweier]

@msweier did you see a new version of shef-parsers that went out yesterday that should fix the empty ts issue?

No i did not. We just got rid of the timeseries as a work around. Glad there is a fix!