[daip] Parallel code in classic aips

[Eric Greisen]

If the compiler can handle it, there are a few Q routines that go
parallel on SGIs and speed things up.  Unfortunately, they are for DFT
component subtraction and gridded is faster than parallel DFT.  If one
could arrange it, simply running 8 IMAGRs, each on a different channel
or range of channels, is the embarrassingly parallel solution.  Of
course, it gets harder to do interaction with the Clean in this case.

Eric