MAG - Metadata updates (SPDF feedback, CAVA plotting)

[jtniehof]

Change Summary

Overview

Metadata and typing update for MAG L2 (and L1d).

• Change quality_flags to UINT1 (8 bits) since it's a single bit; this better matches VALIDMAX
• Add SCALEMIN/SCALEMAX to the field vectors

Closes #3513, #3539. These issues provide more background.

Because #3534 appears to be an issue with old code being deployed, after merge/deploy/reprocess of this, that can probably be closed.

Testing

Unit tests added that directly mirror tests of similar functionality.

[jtniehof]

Rebased to current dev and force-pushed to see if that dealt with the bizarre reduction in coverage of untouched files

[jtniehof]

Incidentally both changes affect burst and norm (or at least should, going to verify on output). EDIT: yep, confirmed, burst is also updated.

[jtniehof]

@alastairtree quick check that -100 to 100 nT is a reasonable default scaling?

[alastairtree]

@alastairtree quick check that -100 to 100 nT is a reasonable default scaling?

+- 100 is probably fine but will make the lines look very flat. When we do field plots we typically adjust the scales based on the data.

[jtniehof]

+- 100 is probably fine but will make the lines look very flat. When we do field plots we typically adjust the scales based on the data.

We always have the possibility to zoom after the fact...this is just a reasonable default for bringing up a large scale plot. Tradeoff between not going off-scale and showing something not-ridiculous. 20 also seems to make sense to me if it's better for you, will just have a little clipping.

[jtniehof]

@bryan-harter : now that all the Tests have completed, the codecov/project has jumped back up to 98+%. ubuntu-latest 3.10 took forever to run and I'm guessing codecov is somehow using that information for its coverage.

[jtniehof]

If someone can poke me once merged/deployed/reprocessed, I need to pull updated samples for SPDF. Thanks!

[lacoak21]

LGTM! Is l1d the earliest product that needs reprocessing?

[jtniehof]

LGTM! Is l1d the earliest product that needs reprocessing?

Should be. I didn't explicitly check that l1c doesn't get affected somehow, but we don't try to plot it in CAVA and we're explicitly casting the flag type here, so don't need to do any changes in earlier levels.

[jtniehof]

l1d is complete; I am reprocessing l2 (norm-srf and burst-srf, then will verify this makes all outputs)