Refactor FastEquals to use SIMD-accelerated SequenceEquals

[Aman-Anas]

This PR updates the FastEquals method to use the SequenceEquals method for faster comparison. The SequenceEquals method is apparently SIMD-enabled and should have better performance, especially for larger struct objects.

I'm currently trying some benchmarks, will update this PR with results when done.

Here are the benchmark results I got:

| Method        | Mean     | Error     | StdDev    |
|-------------- |---------:|----------:|----------:|
| OldEqualsSame | 3.634 ns | 0.0318 ns | 0.0266 ns |
| OldEqualsDiff | 5.434 ns | 0.0333 ns | 0.0295 ns |
| NewEqualsSame | 1.955 ns | 0.0135 ns | 0.0105 ns |
| NewEqualsDiff | 2.182 ns | 0.0734 ns | 0.1730 ns |

This uses a large struct (12 longs) to measure the difference in perf. It's pretty significant the larger your input data.

Here's the source for the benchmark: Benchmark.txt

I also tested with a very small struct (3 ints) but couldn't really find a measurable difference at that size.

[Aman-Anas]

@RevenantX Do I need to add a #ifdef for platforms that don't support Span? I didn't see any in the SpanWriter class, so maybe it's not needed

[RevenantX]

Strange that its faster because when i checked benchmarks SequenceEquals was slower. Maybe depends on structure size or Mono/.Net