This adds:
- LLVM subtarget features to make all the new instructions conditional on
- CPU and FPU names for use on armclang's command line, with default FPUs set so that "armv8.1-m.main+fp" and "armv8.1-m.main+fp.dp" will select the right FPU features
- architecture extension names "mve" and "mve.fp"
- ABI build attribute support for v8.1-M (a new value for Tag_CPU_arch) and MVE (a new actual tag)
With this addition the test runs out of memory on a MIPS board. Shouldn't this loop go through each flag in the Extensions set or at least through every combination of the flags. It seems like overkill to loop from 0 to 16916342 and push mostly the same elements into the vector each time. Even it the platform doesn't run out of memory the test will be unnecessarily slow.