Add ctpop and other bit counting intrinsics

[eholk]

@boggle suggested adding these in a comment on 0a99912#commitcomment-1527123

[eholk]

It'd be nice to have specific use cases for these. Are there programs that we cannot write now that we want to write that would use these intrinsics?

[boggle]

I could use them in the sudoku solver we have as a performance test. Other uses: SAT3 solvers, chess engines (or so I've heard), computing the hamming weight, bitfield based index structures may benefit, and there was a hash table algorithm that uses them.

[auroranockert]

I added a pull-request implementing these intrinsics, if someone wants to have a look: #4245

[graydon]

this landed. Thanks!