google-symptoms missing county data (nov 2022 - feb 2023)

[melange396]

We seem to have no(*) county-level google-symptoms data From November 14, 2022 through February 20, 2023 (*: i didnt do an exhaustive search though, and there is an exception)...

State data seems fine (again, not exhaustively checked). This happens for signals s01-s06 in both raw and smoothed forms, BUT s07 does have data in this period. Maybe google did not have county-level data available at that time but they since filled it in, so we got it when we recently computed the new s07 stuff. ¯\_(ツ)_/¯

[ epivis demonstration ]

I have a vague feeling that this might have been a "known issue", but i cant find it in the signal documentation or in a GH issue in this repo. If it wasnt a known issue, its pretty surprising that we missed it since it was >3 months long and affected so many locations.

We should fix it, and try to figure out how we might alert ourselves to something similar in the future, and potentially try to determine why this happened.

[aysim319]

google_symptom_api_gaps.txt

A quick state does show that compared to the regular runs: 1 for the problem timerange: (November 14, 2022 through February 20, 2023) compared to the ~574 unique geo values is showing for the newly added s07 signal

I started to run patches and it looks like it's coming back with ~1500 fips code for the non s07 signals.
s07 signal is coming a similar ~560 which sounds correct since the length of geo_values were taken from the whole timerange and the number/values for s07 may vary from day to day