Fix #167: double-float-exponent off by one
The following types have a lower bound that is off by one:
single-float-exponent
, double-float-exponent
,
single-float-int-exponent
, and double-float-int-exponent
.
Adjust the lower bound to match what decode-float
and
integer-decode-float
returns for the least-positive-single-float
and least-positive-double-float
.
Add tests for this.