NXsas: review NXcollimator group specification

[prjemian]

The NXcollimator group specification might be good for some neutron sources but is not general. A better choice for some data might be NXaperture (like the recent change in NXcanSAS). The application definition should account for the difference between an aperture (2-D device) and a collimator (3-D device).

[mkoennecke]

The whole NXsas is outdated. The collimator still uses the deprecated NXgeometry way of doing things. In NXdetector and NXsample there is still equatorial_angle which we since have retired in favor of meridional angle. The whole positioning may need review in terms of NXtransformations.