change gsi_obs_space to allow time series to work properly.

[karpob]

Description

This removes the BC_angord_arr_dim in the output dataset, and splits BC_angord array into N number of predictors (BC_angord_1, BC_angord_2, etc). Prior to this fix, the extra BC_angord_arr_dim array dimension relative to JEDI/IODA formats effectively breaks the timeseries plotting capability in eva. It also adds back GsiObsSpace to time_series_utils.py.

What problem does it fix? What new capability does it add?
Fixes problem with GsiObsSpace and timeseries plots. Identified in #215.

Dependencies

If there are PRs that need to be merged before or along with this one, please list them below
None

Impact

Please list the other repositories (if any) that this PR will require changes in (example below)
None

[karpob]

I didn't see the application error test failure originally when I ran the test, but I'm seeing it now on discover. I took the change out, and it's still there. Not sure what's doing this as I haven't touched anything IODA related.

[CoryMartin-NOAA]

@karpob this happens frequently with our EVA CI, I bet it is a python dependency issue.

[karpob]

@CoryMartin-NOAA Thanks. I did a little digging, and that sounds like what it might be, I'm afraid some of the hvplot stuff is a little over my head, so I haven't gotten too far.

[CoryMartin-NOAA]

@karpob I'll try to see what I can figure out in the next day or 2

[CoryMartin-NOAA]

@karpob let's just slap some duct tape on this:
git rm -rf src/eva/tests/config/testIodaObsSpaceAmsuaN19_hvplot.yaml and commit and push, there are other tests for the hvplot interface, there must be something this test does that isn't correct, but I don't think it's worth our time to diagnose right now

[karpob]

Alrighty, pushed it to the branch.