Feature request concerning bitwise operations

[Wen Li]

On x86 processors, I think bitwise operators should be optimized for
individual bit accesses, which should be converted to BT instructions.

For example:

x |= 1<<y; compiles to BTS instead of SHL/OR
x ^= 1<<y; compiles to BTC instead of SHL/XOR
x &= ~(1<<y); compiles to BTR instead of SHL/NOT/AND
if (x&(1<<y))... compiles to BT instead of SHL/AND

Especially when y is not known at compile time, this will provide a
significant performance increase.