Cmm: Remove restriction to x86/x86_64 in MachOp folding
In the Cmm optimizer there is the following comment and code in `cmmMachOpFoldM`:
```
-- powerPC NCG has a TODO for I8/I16 comparisons, so don't try
platformArch platform `elem` [ArchX86, ArchX86_64]
```
The TODO has been done a long time ago and moreover the guard prevents that optimization on all but x86/x86_64 platforms.
issue