Fully retire the 2 NSSP "combined" signals/indicators

[minhkhul]

NSSP has stopped updating the two "combined" signals we collect from them: pct_ed_visits_combined and smoothed_pct_ed_visits_combined. More context here: https://delphi-org.slack.com/archives/C01LZ3A2UMU/p1748756062369789

TODO:

• stop processing/ingesting those signals
• add them to the sircomplainsalot "retired signals"
• mark them as no longer active in the indicators/signals metadata sheet that feeds (among other things) the EpiPortal discovery app

[melange396]

We may want to remove the rows for the "combined" signals that we kept ingesting since october 2024 (when we were writing the same old data that was no longer being updated, but with new "issue" dates). The following SQL might do that, but it is very much untested:

-- '{last_good_issue}' is an epiweek from late sept 2024, something in the neighborhood of '202442'
-- first, make sure this returns the empty set (it should and i will be surprised if not) to make sure we arent throwing away any updated values...
SELECT * FROM eptimetric_latest el JOIN epimetric_full ef USING (`signal_key_id`, `geo_key_id`, `time_type`, `time_value`) WHERE (signal_key_id = 817 OR signal_key_id = 821) AND ef.issue = {last_good_issue} AND time_type = 'week' AND el.value <> ef.value;
-- then delete any issues in the "full" table past the cutoff date...
DELETE FROM epimetric_full WHERE (signal_key_id = 817 OR signal_key_id = 821) AND issue > {last_good_issue} AND time_type = 'week';
-- then set epimetric_latest to the cutoff date (the last issue we kept)...
INSERT INTO epimetric_latest SELECT * FROM epimetric_full WHERE (signal_key_id = 817 OR signal_key_id = 821) AND issue = {last_good_issue} AND time_type = 'week'
        ON DUPLICATE KEY UPDATE
            `epimetric_id` = `epimetric_id`,
            `value_updated_timestamp` = `value_updated_timestamp`,
            `value` = `value`,
            `stderr` = `stderr`,
            `sample_size` = `sample_size`,
            `issue` = `issue`,
            `lag` = `lag`,
            `missing_value` = `missing_value`,
            `missing_stderr` = `missing_stderr`,
            `missing_sample_size` = `missing_sample_size`;