Add SVE support about simdutf

[lemire]

That's a great project. It will take some time, but given that every system with SVE has NEON, one could start with the NEON kernel (called arm64) and create a new one for processors with SVE. A tricky part that I do not know how to resolve at this time would be runtime detection of SVE instructions. It may not be possible to do so... that is, the binary might have a hard requirement on SVE once built.

[clausecker]

I had thought about NEON support before. As NEON does not have an equivalent to the vpcompressb instruction nor anything like pdep or pext, reengineering of the code will be required to make it work. At least the compress operations will have to be emulated somehow.

A tricky part that I do not know how to resolve at this time would be runtime detection of SVE instructions. It may not be possible to do so... that is, the binary might have a hard requirement on SVE once built.

Support for SVE is indicated in ELF auxiliary vectors which can be queried with operating system specific functions.

[bkmgit]

Thanks. Will start on this next week. Tests can be done on Graviton 3, Yitian 710, A64FX and Exynos 2200.