Adds OpenMP to qsort, should also improve test speed a bit

[sterrettm2]

This adds OpenMP acceleration to quicksort, in addition to making some changes to the testing code. On my current testing system, this gives up to a 3x speedup, though I think more may be achieved on a stronger system.

Benchmarks:

10m

Comparing simdsort/random_10m (from /home/sterrettm/simd-sort/x86-simd-sort/.bench/main/builddir/benchexe) to simdsort/random_10m (from /home/sterrettm/simd-sort/x86-simd-sort/.bench/qsort-openmp/builddir/benchexe)
Benchmark                                                                       Time             CPU      Time Old      Time New       CPU Old       CPU New
------------------------------------------------------------------------------------------------------------------------------------------------------------ 50815267
[simdsort/random_10m vs. simdsort/random_10m]/uint64_t_mean                  -0.6462         -0.6461     161374745      57099765     161274843      57081772
[simdsort/random_10m vs. simdsort/random_10m]/int64_t_mean                   -0.6722         -0.6720     161381007      52894903     161242920      52883775
[simdsort/random_10m vs. simdsort/random_10m]/uint32_t_mean                  -0.6683         -0.6684      76346854      25321800      76335248      25316127
[simdsort/random_10m vs. simdsort/random_10m]/int32_t_mean                   -0.6461         -0.6462      74568936      26387521      74555377      26374643
[simdsort/random_10m vs. simdsort/random_10m]/uint16_t_mean                  +0.0286         +0.0286     306422934     315196752     306379370     315156800
[simdsort/random_10m vs. simdsort/random_10m]/int16_t_mean                   +0.0159         +0.0159     308471111     313372843     308432095     313323090
[simdsort/random_10m vs. simdsort/random_10m]/float_mean                     -0.6593         -0.6593      75784101      25821957      75777886      25815941
[simdsort/random_10m vs. simdsort/random_10m]/double_mean                    -0.6613         -0.6613     140408876      47560519     140340074      47530458
OVERALL_GEOMEAN                                                              -0.5520         -0.5520             0             0             0             0

1m

Comparing simdsort/random_1m (from /home/sterrettm/simd-sort/x86-simd-sort/.bench/main/builddir/benchexe) to simdsort/random_1m (from /home/sterrettm/simd-sort/x86-simd-sort/.bench/qsort-openmp/builddir/benchexe)
Benchmark                                                                     Time             CPU      Time Old      Time New       CPU Old       CPU New
----------------------------------------------------------------------------------------------------------------------------------------------------------
[simdsort/random_1m vs. simdsort/random_1m]/uint64_t_mean                  -0.5563         -0.5564      13892838       6163743      13893768       6162738
[simdsort/random_1m vs. simdsort/random_1m]/int64_t_mean                   -0.5439         -0.5441      13943919       6359981      13944683       6358016
[simdsort/random_1m vs. simdsort/random_1m]/uint32_t_mean                  -0.4661         -0.4662       6244794       3334286       6244370       3333435
[simdsort/random_1m vs. simdsort/random_1m]/int32_t_mean                   -0.3862         -0.3861       6172364       3788625       6173740       3789904
[simdsort/random_1m vs. simdsort/random_1m]/uint16_t_mean                  -0.0138         -0.0138      31545637      31108834      31548000      31111137
[simdsort/random_1m vs. simdsort/random_1m]/int16_t_mean                   -0.0076         -0.0076      30964472      30728775      30967064      30730877467       3261958
[simdsort/random_1m vs. simdsort/random_1m]/float_mean                     -0.4568         -0.4566       6418410       3486598       6419908       3488443
[simdsort/random_1m vs. simdsort/random_1m]/double_mean                    -0.5486         -0.5488      11469483       5177166      11469799       5174866
OVERALL_GEOMEAN                                                            -0.4088         -0.4089             0             0             0             0

And to show that there does not seem to be a regression with small sizes due to the SIMD logic:

128

Benchmark                                                                       Time             CPU      Time Old      Time New       CPU Old       CPU New
------------------------------------------------------------------------------------------------------------------------------------------------------------
[simdsort/random_128 vs. simdsort/random_128]/uint64_t_mean                  -0.0442         -0.0512           866           828           878           833
[simdsort/random_128 vs. simdsort/random_128]/int64_t_mean                   -0.0006         -0.0028           821           821           828           826
[simdsort/random_128 vs. simdsort/random_128]/uint32_t_mean                  +0.0152         +0.0150           583           591           587           596
[simdsort/random_128 vs. simdsort/random_128]/int32_t_mean                   +0.0210         +0.0150           582           594           588           597
[simdsort/random_128 vs. simdsort/random_128]/uint16_t_mean                  -0.0034         +0.0008          1933          1926          1939          1941
[simdsort/random_128 vs. simdsort/random_128]/int16_t_mean                   +0.0061         +0.0104          1889          1901          1896          1916
[simdsort/random_128 vs. simdsort/random_128]/float_mean                     +0.0012         +0.0022           599           600           605           606
[simdsort/random_128 vs. simdsort/random_128]/double_mean                    +0.0079         +0.0086           707           712           711           717
OVERALL_GEOMEAN                                                              +0.0003         -0.0003             0             0             0             0

In addition, this adds larger tests for quicksort to test the OpenMP logic. Out of concern for the runtime of the test suite, I modified both these new tests and the older kv tests to only use the larger sizes for the main sort, the only one that uses OpenMP and thus needs them. I believe this should result in a noticeable net reduction in the runtime of the test suite.