Add Partition to SVE micro benchmark

[ylpoonlg]

Add micro benchmark for the partition function in scalar and SVE.

@dotnet/arm64-contrib @a74nh

[ylpoonlg]

@dotnet-policy-service agree company="Arm"

[ylpoonlg]

Performance results (run on Nvidia Grace)

| Method       | Size  | Mean         | Error      | StdDev     | Median       | Min          | Max          | Allocated |
|------------- |------ |-------------:|-----------:|-----------:|-------------:|-------------:|-------------:|----------:|
| Scalar       | 15    |    10.995 ns |  0.2522 ns |  0.2477 ns |    10.844 ns |    10.827 ns |    11.568 ns |         - |
| SvePartition | 15    |     9.756 ns |  0.0081 ns |  0.0076 ns |     9.752 ns |     9.749 ns |     9.772 ns |         - |
| Scalar       | 127   |   100.488 ns |  0.0617 ns |  0.0515 ns |   100.472 ns |   100.419 ns |   100.615 ns |         - |
| SvePartition | 127   |    85.037 ns |  0.0778 ns |  0.0650 ns |    85.032 ns |    84.953 ns |    85.192 ns |         - |
| Scalar       | 527   |   431.728 ns |  0.2506 ns |  0.2093 ns |   431.621 ns |   431.529 ns |   432.207 ns |         - |
| SvePartition | 527   |   357.896 ns |  0.0638 ns |  0.0533 ns |   357.912 ns |   357.796 ns |   357.982 ns |         - |
| Scalar       | 10015 | 8,338.500 ns |  5.8883 ns |  5.2198 ns | 8,338.681 ns | 8,328.850 ns | 8,346.849 ns |         - |
| SvePartition | 10015 | 6,968.263 ns | 11.8289 ns | 11.0648 ns | 6,972.462 ns | 6,932.894 ns | 6,978.159 ns |         - |

[a74nh]

Performance results (run on Nvidia Grace)

| Method       | Size  | Mean         | Error      | StdDev     | Median       | Min          | Max          | Allocated |
|------------- |------ |-------------:|-----------:|-----------:|-------------:|-------------:|-------------:|----------:|
| Scalar       | 15    |    10.995 ns |  0.2522 ns |  0.2477 ns |    10.844 ns |    10.827 ns |    11.568 ns |         - |
| SvePartition | 15    |     9.756 ns |  0.0081 ns |  0.0076 ns |     9.752 ns |     9.749 ns |     9.772 ns |         - |
| Scalar       | 127   |   100.488 ns |  0.0617 ns |  0.0515 ns |   100.472 ns |   100.419 ns |   100.615 ns |         - |
| SvePartition | 127   |    85.037 ns |  0.0778 ns |  0.0650 ns |    85.032 ns |    84.953 ns |    85.192 ns |         - |
| Scalar       | 527   |   431.728 ns |  0.2506 ns |  0.2093 ns |   431.621 ns |   431.529 ns |   432.207 ns |         - |
| SvePartition | 527   |   357.896 ns |  0.0638 ns |  0.0533 ns |   357.912 ns |   357.796 ns |   357.982 ns |         - |
| Scalar       | 10015 | 8,338.500 ns |  5.8883 ns |  5.2198 ns | 8,338.681 ns | 8,328.850 ns | 8,346.849 ns |         - |
| SvePartition | 10015 | 6,968.263 ns | 11.8289 ns | 11.0648 ns | 6,972.462 ns | 6,932.894 ns | 6,978.159 ns |         - |

10% to 20% improvement with SVE over scalar, getting better as the size increases. Not as fast as I was hoping for, but it's a definite gain. We should probably do some comparisons against C++ to see if it's the algorithm/hardware or the CoreCLR codegen.

[a74nh]

@LoopedBard3 @DrewScoggins @adamsitnik - not sure who best to ping for this, so going off who replied on #4841.

@ylpoonlg is with us for a few months, so I've asked him to make some improvements/additions to the SVE tests as we could do with improving the coverage.