cranelift x64: use the LZCNT instruction for Clz when it's available

[bnjbvr]

No description provided.

[cfallin]

Thanks, this should be a nice improvement when the CPU supports it!

Could you add a lowering test with the lzcnt flag set?

Also, if it's not too much trouble, it looks like tzcnt also exists and I see the TODO under ctz below; would you be able to add that as well? I'm fine with doing this in a separate PR if you'd like to get this in first.

[bnjbvr]

@cfallin done! Also done popcnt, same category of easy todos.

For Clz/Ctz, in theory we could use Tzcount/Lzcount for i128 too: that requires local control flow (e.g. for Clz, lzcount on high component, if it's non zero it's the right result, otherwise the result is lzcount on the low component), so a new Vcode inst. Could be done in a subsequent PR.

[cfallin]

LGTM; thanks!