754R Annex D: debugging should be in the debugger

[David G Hough 754R work <754r@xxxxxxxxxxx>]

Joe Darcy commented yesterday that debugging should be in the debuggers,
and we should recognize that not all deployment environments will be good
debugging environments and not constrain the former unnecessarily in 754R.

That suggested to me that somebody should write a 754R
Annex D: Desirable Debugging Devices
that should discuss all the features of 754 that are or would be
especially useful for debugging numerical code.
In my opinion that list would include:

        dynamic modes of all types
        traps generating Unix-style signals that invoke the debugger
        signaling NaNs
        a mode to vary precision (e.g. double rounded to only 52 or only 51
                sig bits)
        methods to vary dynamic modes for specific subprograms