curate format-dates: support date ranges

[joverlee521]

Context

NCBI Datasets has started returning date ranges in the format of [2001 TO 2002], which has caused errors in the ingest workflow during augur curate format-dates (norovirus, rabies, seasonal-cov).

Related to #1494, but different in that this needs to support parsing an existing date range format rather than apply a date range to a field.

Description

curate format-dates should be able to convert the date range to the Augur supported date range format YYYY-MM-DD/YYYY-MM-DD.

[2001 TO 2002] -> 2001-XX-XX/2002-XX-XX 2001-01-01/2002-12-31

[victorlin]

I can take a look at this today (and finally get back to your comments on #1828)

[joverlee521]

Thanks @victorlin!

I'll add --failure-reporting "warn" to the failing workflows which should make this less of a pressing issue. This will mask those failing date ranges to XXXX-XX-XX, which I think is ~fine for now.

[joverlee521]

FYI, I think this will require more changes outside of augur curate to support date ranges through the full pipeline.

I considered using annotations for filling in date ranges, but when I tested the phylo workflow locally, I ran into error with augur filter

ERROR: Invalid date '2001-XX-XX/2002-XX-XX': Date does not match format `%Y-%m-%d`.

[victorlin]

Each end of the date range must be an exact date. It should be safe to append the lower bound with -01-01 and upper bound with -12-31 (that's what I've done in #2001).

[victorlin]

If the PR goes well, I can merge and release today so we don't need to add --failure-reporting "warn" in 3 different places which would need 3 additional PRs to revert later.

[joverlee521]

Each end of the date range must be an exact date. It should be safe to append the lower bound with -01-01 and upper bound with -12-31 (that's what I've done in #2001).

Ahhh, I see. Okay that makes sense!

[joverlee521]

chikv ingest failed this morning as well. No logs because the GH Action workflow doesn't define the artifact paths, but I ran ingest locally and see the same date range error.

[victorlin]

Releasing the fix in 33.3.0.