PNG: Add SSE/AVX version of Sub, Up, Average and Paethfilters

[brianpopow]

Prerequisites

• I have written a descriptive pull-request title
• I have verified that there are no overlapping pull-requests open
• I have verified that I am following the existing coding patterns and practice as demonstrated in the repository. These follow strict Stylecop rules 👮.
• I have provided test coverage for my change (where applicable)

Description

This PR adds a SSE2 version of the average filter (4 bytes per pixel only), which is used for decoding PNG's.
I tried a SSE2 version for 3 bytes per pixel also, but benchmarks did not show god results. Probably due to misaligned read/write.

edit:
PR now also includes SSE/AVX versions of Paeth, Sub and Up filter.

Benchmark results:

master:

BenchmarkDotNet=v0.13.0, OS=Windows 10.0.19043.1526 (21H1/May2021Update)
Intel Core i7-6700K CPU 4.00GHz (Skylake), 1 CPU, 8 logical and 4 physical cores
.NET SDK=6.0.200
  [Host]     : .NET 5.0.14 (5.0.1422.5710), X64 RyuJIT
  Job-OGQEMG : .NET 5.0.14 (5.0.1422.5710), X64 RyuJIT
|                      Method |     Mean |     Error |    StdDev | Gen 0 | Gen 1 | Gen 2 | Allocated |
|---------------------------- |---------:|----------:|----------:|------:|------:|------:|----------:|
| 'Average-filtered PNG file' | 2.626 ms | 0.0179 ms | 0.0046 ms |     - |     - |     - |      3 KB |

PR:

BenchmarkDotNet=v0.13.0, OS=Windows 10.0.19043.1526 (21H1/May2021Update)
Intel Core i7-6700K CPU 4.00GHz (Skylake), 1 CPU, 8 logical and 4 physical cores
.NET SDK=6.0.200
  [Host]     : .NET 5.0.14 (5.0.1422.5710), X64 RyuJIT
  Job-OGQEMG : .NET 5.0.14 (5.0.1422.5710), X64 RyuJIT

Runtime=.NET 5.0  Arguments=/p:DebugType=portable  IterationCount=5
LaunchCount=1  WarmupCount=3

|                             Method |        Mean |     Error |   StdDev | Ratio | RatioSD |  Gen 0 | Gen 1 | Gen 2 | Allocated |
|----------------------------------- |------------:|----------:|---------:|------:|--------:|-------:|------:|------:|----------:|
| 'Average-filtered PNG file (4bpp)' | 2,212.37 us | 14.979 us | 3.890 us | 55.63 |    5.27 |      - |     - |     - |      3 KB |

Testimage was:

[brianpopow]

@gfoidl thanks for the suggestions!

[brianpopow]

I have added SSE/AVX versions of the other filters, too.

Benchmark with random data of size 1024:

BenchmarkDotNet=v0.13.1, OS=Windows 10.0.19043.1526 (21H1/May2021Update)
Intel Core i7-6700K CPU 4.00GHz (Skylake), 1 CPU, 8 logical and 4 physical cores
.NET SDK=6.0.200
  [Host]     : .NET 5.0.14 (5.0.1422.5710), X64 RyuJIT
  DefaultJob : .NET 5.0.14 (5.0.1422.5710), X64 RyuJIT


|      Method |        Mean |     Error |    StdDev |
|------------ |------------:|----------:|----------:|
|    UpScalar |   525.50 ns |  2.733 ns |  2.557 ns |
|      UpSse2 |    36.66 ns |  0.160 ns |  0.150 ns |
|      UpAvx2 |    19.11 ns |  0.074 ns |  0.069 ns |
|   SubScalar |   654.48 ns |  3.637 ns |  3.224 ns |
|     SubSse2 |   172.30 ns |  2.738 ns |  2.561 ns |
| PaethScalar | 6,643.09 ns | 23.076 ns | 21.585 ns |
|   PaethSse2 |   674.23 ns |  2.468 ns |  2.309 ns |