Refactor cdc_covidnet to use geo utils

[chinandrew]

For #306. Depends on #310 (which is currently the base in this PR for easier review).

Summary of changes:

• Convert cdc_covidnet indicator to use the GeoMapper utility.
• Remove old cdc_covidnet specific geo_maps file and code

[chinandrew]

Unrelated to this refactor but I noticed the static_path argument doesn't appear to be used in update_sensor(), is there any reason it's still included? @eujing

EDIT: I lied, not unrelated

[eujing]

@chinandrew static_path was only used with the original implementation for GeoMaps(static_path), but since that has been refactored I dont think its needed anywhere else.
I agree I think it should be removed!

[chinandrew]

@chinandrew static_path was only used with the original implementation for GeoMaps(static_path), but since that has been refactored I dont think its needed anywhere else.
I agree I think it should be removed!

Cool thanks for verifying, will remove.

[sgsmob]

Looks good to me pending an answer on my question above.

[sgsmob]

Is there a test that fails if this line is omitted?

[chinandrew]

So if hosp_df[APIConfig.STATE_COL] = hosp_df["state_id"].str.upper() is omitted, tests fail because APIConfig.STATE_COL isn't the two letter abbreviation.

if hosp_df.drop("state_id", axis=1, inplace=True) is omitted, tests fail because dimensions are wrong.

I also actually just realized column order for these fields doesn't matter since there's a hosp_df.set_index(["date", "geo_id"], inplace=True) right after, so this is actually just to avoid name collisions since there isn't currently functionality to replace a column while keeping the name the same. I've updated the comment to reflect that.