Assembly implementation of 4X1 & 4X2 Huffman for AArch64

[chiranmoyf]

Assembly implementation of 4X1 & 4X2 Huffman Decoder for AArch64.

Benchmarked using fullbench.
Instance: m7g.4xlarge (Graviton 3)
Observation: Gains to ZSTD's decompression speed. Compression speed remained flat.

#	Benchmark	Baseline (MB/s)	ASM (MB/s)	Change (%)
0	compress	254.2	254.3	0.04%
1	decompress	1214.9	1237.1	1.83%
2	compress_freshCCtx	254.1	254.5	0.16%
3	decompressDCtx	1214.8	1231.1	1.34%
4	compressContinue	254.2	254.0	-0.08%
5	compressContinue_extDict	253.6	253.5	-0.04%
6	decompressContinue	1217.6	1226.8	0.76%
7	compressStream	241.2	240.8	-0.17%
8	compressStream_freshCCtx	240.0	241.0	0.42%
9	decompressStream	1217.4	1231.2	1.13%
10	compress2	254.1	254.5	0.16%
11	compressStream2, end	254.4	254.2	-0.08%
12	compressStream2, end & short	253.0	253.3	0.12%
13	compressStream2, continue	241.0	240.4	-0.25%
14	compressStream2, -T2, continue	433.1	433.1	0.00%
15	compressStream2, -T2, end	433.1	432.9	-0.05%
16	compressSequences	589.8	590.4	0.10%
17	compressSequencesAndLiterals	745.5	749.1	0.48%
18	convertSequences (1st block)	25756.0	25721.2	-0.14%
19	get1BlockSummary (1st block)	26810.3	26809.3	0.00%
20	decodeLiteralsHeader (1st block)	154826.3	154323.8	-0.32%

This PR is a joint contribution by:
Abhishek Jain(@abhijain1204fujitsu)
Mohak Goel(@mohakgoel1)