$Id$
float takes a polytope file with rational or integer coefficents,
and outputs an approximately equivelent one with floating point
coefficents.
WARNING: Assumes that numerator and denominator will fit in long integer,
unless compiled with multiprecision support.