英雄帖: Implement XZ BCJ filter for LoongArch

[xen0n]

Good reference materials:

• https://github.com/tukaani-project/xz/blob/master/src/liblzma/simple/arm64.c
• https://github.com/loongson/la-abi-specs/blob/release/laelf.adoc
• https://github.com/loongson-community/loongarch-opcodes -- for precise description of instruction formats; the 9 official formats doesn't cover even lu12i.w or pcalau12i so you would have to come up with something else anyways.

(the RISC-V filter is arguably less suitable for LoongArch as a reference, because while the two architectures' ABIs are similar, RISC-V has compressed instruction formats that LoongArch doesn't.)

Also, you would likely need to benchmark the results to determine if this optimization is worthwhile at all, before submitting things upstream for review.

Please get in touch with us by replying below if you wish to take this.