Refactor bytesutil, add support for go1.20 slice to array conversions

[prestonvanloon]

What type of PR is this?

Other

What does this PR do? Why is it needed?

This PR adds early support for go1.20 slice to array conversions. Also see go1.20 notes

Other changes:

• Refactored this package so that everything wasn't in one bytes.go and bytes_test.go file.
• Fixed PadTo where it would unnecessary pad a slice that was exactly the desired size.
• Added benchmark to show improvement for bytes32 conversions.

Which issues(s) does this PR fix?

Other notes for review

Benchmark results indicate a significant reduction in conversions.

// Before
BenchmarkToBytes32-8    1000000000               0.4843 ns/op                                            

// After
BenchmarkToBytes32-8    1000000000               0.2438 ns/op

Tested with gotip @ go version devel go1.20-db36eca Thu Dec 29 23:36:05 2022 +0000 linux/amd64

This is a very high traffic method in Prysm and performance improvements could come in during epoch processing and other high impact workloads.

[saolyn]

Had a look through everything and it looks like a huge improvement.
I did notice there are a couple of functions in bytes.go ( FromBytes48Array ) and a couple others in integers.go that don't seem to have any unit tests. Might be worth adding them as well.

[prestonvanloon]

Thanks @saolyn. I don't have capacity to add those missing tests at this time. They only notable additions in this PR are in the file bytes_go120.go. Everything else is a refactoring of existing stuff.