feat(asm): add VEX shuffle, extract, permute, move and FP forms

Assisted-by: Qwen 3.8 Max Preview

asm/instrs.go

@@ -136,6 +136,17 @@ func (e *enc) encodeALU(op struct {
 		return e.encodeALUImm(op.digit, dst, int64(imm), size)
 	}
 
+	// CMP accepts the immediate in the second position too — CMPL CX, $31 is
+	// the form the Go assembler itself accepts — and encodes it identically
+	// (CMP r/m, imm sets the flags as first − second).  No other ALU op takes
+	// an immediate destination.
+	if imm, ok := dst.(Imm); ok {
+		if op.digit != 7 {
+			return fmt.Errorf("immediate must be the source operand")
+		}
+		return e.encodeALUImm(op.digit, src, int64(imm), size)
+	}
+
 	dstReg, dstIsReg := dst.(Reg)
 	srcReg, srcIsReg := src.(Reg)
 	switch {