- solidify/overlap.c (overlap): added speed testing loop
- solidify/overlap.c (zmix): instead of floor/ceil, we just use
floor/floor+1. With the ramp() change below, an 18% speed increase results
- solidify/overlap.c (ramp): after the above change, w0 and w1 can never be
zero, so we don't have to test for this case
- solidify/array.h (get_bounded), solidify/array.c: inline get_bounded(),
yielding a 13% speed increase