DM2 session notes: STC and Source

Arnold presented STC status and Jonathan presented
an abstract data model of the STC without XML-related optimizations.
This latter model was well received; but it needs more work
to clean it up.

For STC, the group suggested that
 - more examples were needed, with annotated explanations
 - the documentation amended to loudly indicate that RedshiftAxis
   also includes Doppler axes (i.e. velocity) to keep the
   galactic community less unhappy
 - the document needs to be completed and circulated as soon as
   possible, but then it should go forward.
 - the status of implementations was dicussed: Dave Berry's AST 
   library, JHU footprint service, VOEvent, Registry.

Inaki presented the status of the Source Data Model and
illustrated an attempt to map it to Spectrum and STC utypes.
He pointed out a problem: in the Source model, 
there are fields for both Luminosity and Flux  (or something
like that). But there is only 1 utype in Spectrum for the
spectral Y-axis, i.e.  Spectrum.Flux.Value  (or something like that).

Inaki and Pedro had suggested adding extra utypes in the model
corresponding to all possible different UCD cases that
Spectrum can represent, thus Spectrum.Flux.Luminosity,
Spectrum.Flux.Counts, etc. etc.. many times.

Jonathan said that the Spectrum model describes a single view
of a spectrum with one choice of x-axis and one choice of y-axis.
So what Inaki has is really references to TWO instances of Spectrum
in his model. This points out that we need ways to refer to this
using utype. Jonathan suggested that we have a way to declare
multiple instances, e.g.
  declare LSpectrum is an instance of Spectrum  (for the luminosity vs lambda view)
    with LSpectrum.Flux.ucd = phot.lum
  declare FSpectrum is an instance of Spectrum  (for the flux vs lambda view)  
    with LSpectrum.Flux.ucd = phot.flux
and then use
  LSpectrum.Flux.Value and FSpectrum.Flux.Value as utypes for the
two cases. This is much more economical, and logical, than
adding extra utypes for all possible ucd values of the things in the
model that can take different value. But we'll have to figure out
the right syntax and mechanism to do this.

We noted that the effort on the catalog container model had been
set aside, and should be taken up in conjunction with the VOQL related
table access protocol.