bignums and native back end

[Henry Cejtin]

Is there any thing I can do to get bignums working with the native back end?
Aside from a few primitives, I would think that the only real thing is a bit
of special analysis to handle the cases where one of the arguments is known
to be small (things like `x + 1' or `x = 0').
Note, there is a new version of the GMP GNU multi-precision code which I have
grabbed.  It doesn't look like it will be anything more than a trivial change
to switch to it and it claims to be faster.