Implement search-skip optimization framework for ZSTD compressors

[BiplabRaut]

• Add ZSTD_ENABLE_SEARCH_SKIP_HIGH macro to portability_macros.h for conditional compilation
• Extend compression level parameters in clevels.h with search-skip configurations
• Integrate adaptive search skipping in zstd_fast.c, zstd_double_fast.c, and zstd_lazy.c
• Update build configuration in libzstd.mk to support new optimization flags
• Optimize compression speed through strategic match search reduction with minor compression ratio tradeoff

Performance Results

Test Setup: AMD Genoa (Zen4), SMT OFF, GCC 14.2, single-threaded. ~2% run-to-run variation expected.

Level 3

Dataset	Baseline CS (MB/s)	Optimized CS (MB/s)	CS Speedup	Ratio Baseline	Ratio Optimized	Ratio Change	DS Change
silesia.tar	299.1	319.3	+6.8%	3.206	3.165	-1.3%	+2.6%
calgary.tar	246.1	258.8	+5.2%	3.107	3.095	-0.4%	+3.6%
canterbury.tar	306.3	309.5	+1.0%	4.461	4.468	+0.2%	+2.0%
freeBSD-13.2	698.5	876.8	+25.5%	1.598	1.588	-0.6%	-0.6%

Level 5

Dataset	Baseline CS (MB/s)	Optimized CS (MB/s)	CS Speedup	Ratio Baseline	Ratio Optimized	Ratio Change	DS Change
silesia.tar	152.1	258.7	+70.1%	3.378	3.302	-2.2%	+0.7%
calgary.tar	131.1	214.1	+63.3%	3.220	3.179	-1.3%	+4.2%
canterbury.tar	173.2	261.8	+51.2%	4.577	4.539	-0.8%	+3.8%
freeBSD-13.2	555.2	763.5	+37.5%	1.605	1.599	-0.4%	-2.0%

Level 6

Dataset	Baseline CS (MB/s)	Optimized CS (MB/s)	CS Speedup	Ratio Baseline	Ratio Optimized	Ratio Change	DS Change
silesia.tar	111.5	153.6	+37.8%	3.462	3.375	-2.5%	-6.4%
calgary.tar	94.3	133.7	+41.8%	3.309	3.218	-2.7%	-6.9%
canterbury.tar	111.2	178.6	+60.7%	4.708	4.580	-2.7%	-5.8%
freeBSD-13.2	421.3	560.6	+33.1%	1.610	1.604	-0.4%	-3.1%

Summary

• Level 3: 1-26% compression speed gain with negligible ratio impact
• Level 5: 37-70% compression speed gain with 0.4-2.2% ratio cost
• Level 6: 33-61% compression speed gain with 2.5-2.7% ratio cost and 3-7% decompression regression
• Best suited for applications where compression throughput is the primary requirement