remove dependency on primary index by converting queries that are agiqanst the id field to gets.

[randytpierce]

remove dependency on primary index by converting queries that are agiqanst the id field to gets. These are queries for job documents which aren't even all that large. By changing them from queries to data fetches they are a lot more efficient and don't require either a special index or a primary index.

[github-actions[bot]]

Package	Line Rate	Branch Rate	Health
vxingest	33%	38%	❌
vxingest.builder_common	27%	15%	❌
vxingest.ctc_to_cb	12%	1%	❌
vxingest.grib2_to_cb	13%	1%	❌
vxingest.netcdf_to_cb	13%	1%	❌
vxingest.partial_sums_to_cb	12%	1%	❌
vxingest.utilities	30%	32%	❌
Summary	16% (490 / 3033)	6% (40 / 712)	❌

[randytpierce]

This is really strait-forward and should be covered by the integration tests. NOTE I could have also changed ...

Select id as ingest_document_ids
from `vxdata`._default.METAR
where meta().id = "MD:V01:METAR:obs:ingest:netcdf"

to

Select id as ingest_document_ids
from `vxdata`._default.METAR
USE KEYS ["JOB:V01:METAR:CTC:CEILING:MODEL:OPS"]

but I think by-passing the query processor is probably more efficient.