Replace std::array with std::span in roughtime hashNode function

[KaganCanSit]

This optimization was spotted by @reneme in review PR #4783. While there are more instances of `typecast_copy' in this codebase, I wanted to improve it with a small change for this part only.

Referenced review comment: #4783 (comment)

Quote:

General remark (please don't tackle in this pull request): This roughtime implementation makes heavy use of std::array<> in places where (nowadays) it would be better off to use statically-sized std::span.

Here, hashNode is a good example: if it would be implemented using spans, you could avoid the data copies and simplify the code:

   BufferSlicer slicer(path);
   for(size_t level = 0; level < levels; ++level) {
      hashNode(hash, slicer.take<64>(), index & 1);
      index >>= 1;
   }

There are likely other places in this implementation that could avoid the typecast_copy (which is just a fancy name for an unnecessary data-copy).

Following the merge of this branch, I will need to rebase #4783 against master. This should not present any issues. Please let me know if you require any modifications.

Best regards.

[randombit]

LGTM thanks

[KaganCanSit]

LGTM thanks

Thanks for the quick response @randombit. I'm on Github today and will generally make time, but if you have time and see fit, I'd also love to complete and merge enhancement #4783 🤝.

Nice day.

[coveralls]

coverage: 90.626% (+0.004%) from 90.622%
when pulling d9e1c67 on KaganCanSit:reduce-copy-for-rougtime-file-with-span
into e18675f on randombit:master.