[llvm] [SystemZ] Add a SystemZ specific pre-RA scheduling strategy. (PR #135076)

[Jonas Paulsson via llvm-commits]

JonPsson1 wrote:

These are the latest real bencharking results, which show that this seems to have good impact on the spec-FP.

```
f503.bwaves_r                    818.26s             817.67s                   -0.07%
f507.cactuBSSN_r                 323.14s             282.89s                  -12.46%
f508.namd_r                      208.29s             204.09s                   -2.02%
f510.parest_r                    505.40s             503.92s                   -0.29%
f511.povray_r                    568.38s             566.56s                   -0.32%
f519.lbm_r                       228.51s             222.96s                   -2.43%
f521.wrf_r                       411.52s             411.08s                   -0.11%
f526.blender_r                   267.38s             257.95s                   -3.53%
f527.cam4_r                      306.89s             308.40s                    0.49%
f538.imagick_r                   371.98s             324.00s                  -12.90%
f544.nab_r                       356.02s             344.27s                   -3.30%
f549.fotonik3d_r                 333.67s             333.22s                   -0.13%
f554.roms_r                      219.66s             219.54s                   -0.05%
i500.perlbench_r                 307.38s             308.42s                    0.34%
i502.gcc_r                       169.78s             169.24s                   -0.32%
i505.mcf_r                       283.45s             282.39s                   -0.37%
i520.omnetpp_r                   210.89s             211.89s                    0.47%
i523.xalancbmk_r                 180.09s             181.13s                    0.58%
i525.x264_r                      192.52s             193.03s                    0.26%
i531.deepsjeng_r                 257.37s             257.22s                   -0.06%
i541.leela_r                     478.44s             480.84s                    0.50%
i548.exchange2_r                 312.95s             311.87s                   -0.35%
i557.xz_r                        272.35s             269.74s                   -0.96%
                                                                               ------
--- total time                  7584.32s            7462.32s                 -122.00s
--- geometric mean                                                             -1.68%

```

https://github.com/llvm/llvm-project/pull/135076