Fix spc nonpublic data loading

[STBadman]

Fix, simplify, and make more transparent the logic for loading non-public SWEAP/SPC data

New logic is

1. try downloading from sweap server looking for files with a "psp" prefix
2. If not enough files downloaded, raise an Exception as such and try looking for remaining files with a "spp" prefix
3. If no files found, raise a further Exception try one more time in the public directory (spdf).

This also enforces that the try/except handling checks the error is the one raised on purpose (for any future debugging) and prints the error message, handles the case when the user calls a trange with datetimes rather than strings.

I also removed the if/else statement for last_version to simplify things - I couldn't see how it was necessary as last_version was being passed to the download function in all cases.

[jameswilburlewis]

One of the unit tests failed for this change:

Traceback (most recent call last):
  File "/home/runner/work/pyspedas/pyspedas/pyspedas/projects/psp/tests/tests.py", line 22, in test_unpublished_data
    spc = pyspedas.projects.psp.spc(trange=['2018-11-5', '2018-11-5/06:00'], username='hello', password='world')
  File "/home/runner/work/pyspedas/pyspedas/pyspedas/projects/psp/spc.py", line 113, in spc
    return load(instrument='spc', trange=trange, datatype=datatype, level=level, suffix=suffix, prefix=prefix,
        get_support_data=get_support_data, varformat=varformat, varnames=varnames, downloadonly=downloadonly,
        notplot=notplot, time_clip=time_clip, no_update=no_update, username=username, *** last_version=last_version, force_download=force_download)
  File "/home/runner/work/pyspedas/pyspedas/pyspedas/projects/psp/load.py", line 327, in load
    trange_temp = [remote_dates[len(files)],t_end]    # set new trange to check for spp prefixes
                                            ^^^^^
UnboundLocalError: cannot access local variable 't_end' where it is not associated with a value

t_end isn't defined here -- it should be probably be replaced with trange[1] in line 327 to match the previous code. I think I'll go ahead and merge as-is, then fix the bug and make sure the tests pass. You'll probably want to propagate that change back to your fork.