switch and Math.nan

[Henrik Grubbstr�m (Lysator) @ Pike (-) developers forum]

>I think we should (1) issue a warning when Math.nan is used inside a
>case and (2) make sure that switch (Math.nan) does not match any case.

I believe that the compiler should:

  - Special case NaN if there's an explicit case-label with NaN.

  - Map NaN to the default label (if any) otherwise.

This should be rather simple to do with a minor modification of
switch_svalue_cmpfun().

        /grubba