Issue #76: Boolean expressions in conditional statements

[morganthomas]

Closes #76 .

These changes save exactly 1,000 cycles for the 500 XMSS aggregation test, while reducing bytecode size by 1 and leaving memory usage the same.

These changes make it the XMSS aggregation verifier's responsibility to ensure that the bitield values in the public input are in fact bits.

Unfortunately, the most frequent usage of the if X == 1 idiom in the XMSS aggregation test were not apparently in a context where X can be assumed to be boolean. This is in the condition if merkle_are_left[h] == 1. Changing this to if merkle_are_left[h] resulted in an increased cycle count. Changing this to if !!assume_bool(merkle_are_left[h]) resulted in a more substantially decreased cycle count, but I expect it also would give up soundness.

These changes increase the cycle count by 12 for the WHIR recursion test, while leaving bytecode size and memory usage the same.

I don't know why these changes increase the cycle count by 12 for the WHIR recursion test. I did find usages of the if X == 1 idiom in the WHIR recursion test where X could safely be assumed to be boolean, but for whatever reason, replacing these instances with if !!assume_bool(X) resulted in an increased cycle count.