p3-miden absorption follow-ups

This is looking great! There might be some follow ups we want to address to make things a bit more uniform.

The transcript and stateful hasher crates could probably be absorbed into a common Miden hash module. This was again an artifact from the copied Plonky3 architecture, and the separation makes less sense now.

We should look into defining some classic STARK configurations of hashes, since it's sometimes a little hard to get right (wrapping a stateful-truncating-serializing-hasher can be done incorrectly). I think we'd expose default configurations like Blake3-196/Keccak/Poseidon1/2 (with an explicit packed/unpacked configuration).

I'm thinking that the transcript crate could be in the stark module directly, since it doesn't seem like it would be used anywhere else in the near future.

We might want to reconsider how the benching binary fits in the overall picture, but this is not super important. I think they are useful as examples and quick local regression tests if working on the prover. The real insights will come from e2e runs of the Miden VM on real programs.

_Originally posted by @adr1anh in https://github.com/0xMiden/crypto/issues/935#issuecomment-4205351117_