IMPORTANT: q.rd, README, and bin/ are currently *both* in the GAVO SVN and at
https://github.com/aramknyazyan/ArVO_DaCHs.git It is currently being
synchronised between the two (if at all) by hand.

These are the full plate scans of the DFBS.  They were scraped off a
machine in Trieste that, in turn, was a clone of the machine on which
DFBS extraction was done.

The original files had various problems; in particular the astrometric
calibration still was DSS-style, and many headers had non-ASCII in them.
The local bin/addstandardheaders.py fixes that as far as possible, also
adding metadata from WFPDB.

The WFPDB doesn't have all plates; for the missing ones, there's
data/fallback_meta as provided by the Italian group who did the
extraction (but it's less complete).

Quite a few FITSes are still broken:

* fbs0584, fbs0587, fbs1165, fbs1187, fbs1824M didn't receive a
	header, it seems.  Recovering them might be possible if it were
	found worthwhile.
* fbs0323, fbs1825M, fbs1828M, fbs1830M, fbs1831M, fbs1832M, fbs1834M,
  fbs1834_, fbs1835M, fbs1836M, fbs1837M, fbs1838M, fbs1839M, fbs1840M,
  fbs1841M, fbs1842M, fbs1875, fbs1882, fbs1883, fbs2064, fbs2231, fbs2234,
  fbs2235, fbs2237, fbs2242, fbs2243, fbs2244, fbs2247 are not in WFPDB.
  If they were entered there, they'd probably just work.
* fbsanon+38_cor -- who knows?
* there are extra plates that are either failed observations or anyway
  not objective prism exposures, and that are missing in WFPDB.  There's
  some metadata on them in dfbs-supplement.csv that would need to be
  applied.

Since some of them are so broken that the FITS grammar rejects them, I've moved
them out of the way (to data/broken in Heidelberg).  There's still a trigger in
the q.rd only accepts plates with updated headers.  It'll know them by the
presence of the WFPDB-ID header.


The test set is fbs0017_cor.fits in order to enable test across dfbs
and dfbsspec's test set.