1.0.30.5: optimize some floating point operations

1.0.30.5: optimize some floating point operations

 * Convert (/ <float> <one>) to (+ <float> <zero>), and similarly for *.

 * Convert (/ <float> <minus-one>) to (+ (%negate <float>) <zero>), and
   similarly for *.

 * Convert (* <float> <two>) to (+ <float> <float>).

 * Iff FLOAT-ACCURACY is zero, convert (+ <float> <zero>) and (- <float> <zero>)
   to <float>.