Update Parallel Suffix

[jamierpond]

----------------------------------------------------------------------------------------------
Benchmark                                                    Time             CPU   Iterations
----------------------------------------------------------------------------------------------
runCompressions<4, UseSWAR>                               1318 ns         1291 ns       541838
runCompressions<4, UseOldParallelSuffix>                  7044 ns         6934 ns       100607
runCompressions<4, CompareNewAndOldParallelSuffix>        8651 ns         8514 ns        82220
runCompressions<8, UseSWAR>                               7271 ns         7149 ns        96997
runCompressions<8, UseOldParallelSuffix>                 15037 ns        14796 ns        47364
runCompressions<8, CompareNewAndOldParallelSuffix>       20979 ns        20657 ns        33887
runCompressions<16, UseSWAR>                             12322 ns        12126 ns        57702
runCompressions<16, UseOldParallelSuffix>                27133 ns        26641 ns        26617
runCompressions<16, CompareNewAndOldParallelSuffix>      37861 ns        37270 ns        18767
runCompressions<32, UseSWAR>                             19148 ns        18784 ns        37240
runCompressions<32, UseOldParallelSuffix>                44419 ns        43650 ns        15744
runCompressions<32, CompareNewAndOldParallelSuffix>      62413 ns        61247 ns        11397
runCompressions<64, UseSWAR>                             26320 ns        25932 ns        27007
runCompressions<64, UseOldParallelSuffix>                67408 ns        66484 ns        10566
runCompressions<64, CompareNewAndOldParallelSuffix>      91202 ns        89833 ns         7829

[thecppzoo]

Nice!
You doubled the speed of the SWAR PEXT.