STC ObservationLocation

[Jonathan McDowell]

> will be a formal request for comment period when STC is a PR
If this counts as a minor change which doesn't throw the PR back
to the WG, ok... but really I would prefer if we could
reach consensus prior to the official PR stage.

Sorry Arnold, but I would also vote for the ObsLoc to
be optional for coordinate systems which have already
been corrected for observatory location. In the language
of the Observation DM, the ObsLoc in those cases has become
Provenance (check it if you don't trust the data) rather than
Characterization (what you still need for further work if you
do trust the data). One could have another STC instance for
the original as-observed coordinate system in the Provenance
(optionally!) but the main and required STC would not need
ObsLoc if everything is barycenter corrected.

Roy and others, note however that some data has some
things corrected (e.g. velocity frame) and not others
(apparent position? time?) and so you have to be a little
bit careful about throwing away ObsLoc. But if the 
accuracy of the data is such that remaining corrections
are small compared to the quoted errors, then I think there's
no problem (e.g. if your time accuracy is 10 minutes, who
cares if we are in TT or TDB with a 2 ms max difference...).
This is the sort of thing that should go in a usage guide.

 - Jonathan