ELFIN L2 EPD CDFs not ISTP-compliant

[jameswilburlewis]

I had to increase the strictness of some of the ISTP compliance checks in cdf_to_tplot and store_data, to address some problems in other data sets like Cluster and MMS. This has revealed some issues that were probably always there, but silently failing or working with incorrect data, and now they're generating warnings.

The ELFIN L2 EPD CDFs have some problems with DEPEND_1 attributes having the wrong size for their corresponding data dimension. Here is output from test_epd_l2.py:

21-Jul-24 11:34:47: Downloading remote index: https://data.elfin.ucla.edu/ela/l2/epd/fast/electron/2022/
21-Jul-24 11:34:47: File is current: elfin_data/ela/l2/epd/fast/electron/2022/ela_l2_epdef_20220828_v01.cdf
21-Jul-24 11:34:47: Unable to get ydata for variable ela_pef_Tspin
21-Jul-24 11:34:47: Variable ela_pef_hs_epa_spec DEPEND_1 attribute ela_pef_energies_mean has length 16, but corresponding data dimension has length 10. Ignoring.
21-Jul-24 11:34:47: Variable ela_pef_fs_epa_spec DEPEND_1 attribute ela_pef_energies_mean has length 16, but corresponding data dimension has length 18. Ignoring.
21-Jul-24 11:34:48: ELFIN EPD: LOADING END.

The tests are still passing, but whatever processing is being done with the two affected variables is probably incorrect, since the 'v' or 'v1' data is no longer being passed to store_data.

The issues with ela_pef_energies_mean are also picked up by skteditor.

If reprocessing is inconvenient, this could probably be fixed with a master CDF that has the correct dimension and values for.ela_pef_energies_mean. (Actually, two variables would be needed, because one parent variable has length 10 in that dimension, and the other has length 18.)

By the way, skteditor also complains about the global "Source / Spacecraft Name" attribute, though I don't see exactly what the issue is -- maybe a non-printing character in the string?: