Add functions for performing conversion of 8- and 16-bit integers to floats.
Since the ranges of all these integer types lie within the what can be exactly
represented by both single- and double-precision floating-point values, these
conversions always succeed.
Add a note about the float/1 function.
library/float.m:
Add the new conversion functions.
Note that the behaviour of float/1 is currently undefined if the
argument is outside the range that the floating-point type can exactly
represent integers.
Fix section headings.
tests/hard_coded/Mmakfile:;
tests/hard_coded/float_conv.{m,exp}
A test of int->float conversion; currently for 8- and 16-bit integers.
