[CM] clm changes

[Bill Schottstaedt]

 > $ tkdiff clm.c
 > will show you any differences between your local copy and the version in the repository.

and there have been several lately -- a new "average" generator,
a new argument to make-rand and make-rand-interp to give any
desired random number distribution, the old sum-of-sines is
now sine-bank, and there's a new sum-of-sines paralleling
sum-of-cosines (which itself changed slightly to reduce DC).
I'm still trying to decide the "right way" to add all-pass
delay line interpolation.

These changes are also in the common lisp CLM.  I was surprised
that the fm violin can sound noticeably different if I use (say)
gaussian noise -- something I had never tried before.  (This
is also the case in the physical modeling instruments like
pluck and flute).