This is a little survey looking for flare stars taken by Milcho Tsvetkov
and collaborators from 86 to 91.

Most of the images are multiple exposures representing time series.  These
are fairly beastly to get astrometry for, and hence astrometry is only
available for a few "reference" frames, to whiche the others are bound
through datalink.

The data was delivered to us by Milcho on a portable hard disk in Oct 2017;
this was in a fairly funky hierarchy.  What we did was, with raw_data the
content of Milcho's hard drive:

cp raw_data/*/*_FITS/*.fits data/plates/
cp raw_data/*/*_env/*.jpg data/static/envelopes/
cp raw_data/*/*_jpg/*.jpg data/static/photos/
cat raw_data/*/*_WFPDB/*notes.txt > data/metadata/ESO040notes.txt

However, we had to rename a few mislabeled files, so don't do this if
you don't absolutely have to.

There's bin/make_test_set.py that generates data for the regression tests.

Remove raw_data when we're satisfied all files have been pulled (some
time in 2018 I'd say).