amd64: support vector broadcast and extract instructions

[zhouguangyuan0718]

What changed

Add generic amd64 lowering for vector broadcast, bitwise XOR, ternary XOR, vector GF(2) affine, and 64-bit lane extraction forms used by optimized compression dependencies:

• PEXTRQ (register and memory destinations)
• VBROADCASTI128
• VBROADCASTF32X2
• VBROADCASTSD
• VXORPD
• YMM forms of VPTERNLOGD (XOR form, immediate 0x96)
• YMM forms of VGF2P8AFFINEQB

Add focused translation tests for the new operand forms.

Why

LLGo's real Plan 9 assembly translation reported unsupported amd64 instructions in:

• github.com/minio/crc64nvme/crc64_amd64.s (PEXTRQ)
• github.com/klauspost/reedsolomon/galois_amd64.s
• github.com/klauspost/reedsolomon/galois_gen_amd64.s

The lowering stays in the generic amd64 vector implementation and does not depend on package paths or individual callers.

Validation

• GOWORK=off go test ./...
• go test ./internal/plan9asm in LLGo with the local plan9asm workspace
• Real LLGo translation passed for work/compress/s2, crc64nvme, reedsolomon, and the nopshufb generated variant.

[codecov]

Codecov Report

❌ Patch coverage is 95.65217% with 6 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
amd64_lower_vec.go	95.65%	3 Missing and 3 partials ⚠️

📢 Thoughts on this report? Let us know!