Ensure that ephemeris data are accurate

[taldcroft]

Currently the cheta sync version of ephemeris data is a bit broken:

• Full resolution includes duplicate data where time jumps backward [1].
• 5-minute stats don't go into the future, I think.

[1]: When a new ephemeris file is generated, it overlaps with existing old data. The older data are set to have bad values so that fetch.Msid does the right thing. But cheta sync just goes forward and there is currently no way to backfill the bad values.

One potentially easy solution is to rsync the ephemeris data directories by putting them into ska_sync and not sync via cheta_sync.

I know @matthewdahmer has had some experience in this regard, and I ran into a problem today.

[taldcroft]

Maybe this is overtaken by #257 to allow access to the STK ephemeris. This is a better product anyway and matches what the FOT generally use.

A concern is the required OCC network access. One idea is a hybrid where the compressed STK ephemeris files are stored directly in $SKA/data/eng_archive/data/ephem_stk and those get rsynced. A review tool on HEAD would only use the flight data already on disk.