aarch64: Speedup after vectorizing quant/trellis-cabac using SVE intrinsics and converting to ASM.

This patch vectorizes trellis-cabac processing for AARCH64 with SVE2, providing some perf benefit. Based on SVE intrinsic patch from @nekobasu rejected by community because it contained intrinsics, now converted to pure AARCH64 assembly.