Optimize secret masking, c3id genertion, `marvin`, and checksum validation.

[jdraaijer-microsoft]

This PR improves the secret_masker_test with checksum benchmark by about 32%, from 1300 ns to 880 ns per iteration, and the without checksum by about 15% from 879ns to 734 ns per iteration, on my machine.

It applies the following changes:

• marvin
  • Made more idiomatic by relying on slices to provide us with lengths and offsets (~20% speedup). Updated existing impl to use that after validating slice constraints.
  • Applied rustfmt
  • Increased test coverage by adding offsets and non-full-length tests
• Cross company IDs
  • Decreased amount of work required by removing some unnecessary formatting calls and pre-allocating data for known-length strings.
• Secret masking
  • During validation, skip validating the correctness of the key: if it contains invalid bytes, the signature check will fail and we panic either way, so the error message is lost. If we do want this, we should add it as a "post" step, where we check the validity of the signature bytes in the case that validation fails and report it to the caller. It may be that I misunderstood the point of this check, so please do leave feedback in case it is necessary :)
  • Don't calculate the cross-company ID if we don't need it
  • Use the last() detection to track the current one instead of keeping separate state + cloning and unwrapping a bunch
• Scanning
  • API BREAKING (but functionality remains the same): ScanMatch: use Cow<'a, str> instead of String to avoid unnecessarily copying utf8 data to a new string.
• identifiable secrets
  • Make a bunch of statics const.
  • Remove checks that are expensive on the happy path and have minor perf improvement for the un-happy path, that are not explicitly reported.
  • Decrease amount of en- and decoding required for validating a key signature by re-using existing bytes.

This may be (far too) much for a single PR, but I have tried my best to split the changes into concise commits and am willing to make separate PRs for them if that is preferred.

[jdraaijer-microsoft]

I've updated the commit messages to be more descriptive, and make it obvious which commits touch which parts.

[suvamM]

@jdraaijer-microsoft This PR has some good changes. But reviewing this is hard for me. Could you please scope the changes into smaller PRs?

[jdraaijer-microsoft]

Closing in favor of smaller (future) PRs: #69, #70