Optimize allocations via ArrayPool/MemoryPool in Crc64HashAlgorithm.cs and other places

[Mrnikbobjeff]

Description

Some places can benefit from using a memory arena. I explicitely tested this behaviour on the ComputeHashString implementation of Crc64HashAlgorithm.cs. I noticed that in this instance the benefit only materializes over repeated uses of the array pool. Furthermore, one can use a separate allocation strategy for small inputs where the value may be allocated on the stack in case of the CRC algorithm. I already benchmarked the difference, but instead of just publishin the change in one instance investigate other occurances where one would benefit from using the Memory/ArrayPool.
This is a recreation of a previous issue created with the wrong account
ConsoleApp2.CRC64_Benchmark-report-github.md
The benchmark shows that i reach approximately the same runtime as the original algorithm (I am 1% faster) but depending on the input size I allocate 3 orders of magnitude less than the naive version.

Public API Changes

none

Intended Use-Case

Reduced memory usage