[fitswcs] update of Paper III

[Eric Greisen]

David Berry mailed me this ages ago:

     I'm in the process of adding support for spectral coordinate frames
to the Starlink AST library, and I've consequently been reading through
FITS-WCS paper III. I'm a bit confused as to how the source position
should be specified. Whilst I must confess to being new to spectroscopy,
my understanding is that the source position is needed to allow conversion
between reference frames (the "direction vector" mentioned in the second
sentence of section 7). Is the intention that all spectral WCS headers
will be 3 (or more) dimensional (i.e. one spectral axis and two spatial
axes)? If this is the case, it may be helpful to say so explicitly in the
paper. If this is not the case (i.e. if 1-dimensional, purely spectral,
WCS headers are allowed), then how should the source position be specified
- are more parameters needed in table 9?

The answer is no - no more parameters are need.  Instead the full WCS
nomenclature of Papers I and II should always be used.  Since this was
not obvious to David I decided to add a paragraph to the paper - I do
not know if the wording is good or where exactly it should go.  At
present it is on page 22 "real-world complications".  See revision

        http://www.aoc.nrao.edu/~egreisen/scs_050520.ps.gz
        http://www.aoc.nrao.edu/~egreisen/scs_050520.ps
        http://www.aoc.nrao.edu/~egreisen/scs_050520.pdf

in order of increasing size.

Suggestions?  arguments?

Eric Greisen