Set up validation for all indicators

[sgsmob]

Indicators:

• changehc
• claims_hosp
• doctor_visits
• fb_survey
• google_symptoms
• hhs_hosp
• jhu (Add JHU validation parameters #752)
• nchs_mortality
• quidel_covidtest
• safegraph_patterns
• usafacts

Inactive Indicators:

• cdc_covidnet
• combo_cases
• google_health
• hhs_facilities
• quidel
• safegraph

This should basically involve:

1. Set up bare-bones params.json that specifies:

• global.data_source (the indicator name)
• global.span_length = 14
• dynamic.smoothed_signals
• dynamic.max_expected_lag
• dynamic.min_expected_lag

1. Get the indicator output by running it or downloading some cached files from somewhere
2. Run the validation from the indicator directory using python -m delphi_validation
3. Investigate the output and use it to tune the remaining validation parameters

[qx-teo]

Validation parameters have been added (and implemented) for all but one indicator ran using automation.

Exceptions include:

• nchs_mortality: Indicator provides weekly data, which requires modification of the validator (see Adapt Validator for Weekly Data #1117)
• cron indicators: doctor-visits, fb-survey, hospital-admissions, indicator-combination. Of all these, params.json.template files have been created, but validation isn't currently implemented in the pipeline yet.

[nmdefries]

Validation is rolled out to all daily signals in dry-run mode. Acquisition blocking is turned off because validation is too sensitive for it to be useful.