chore: update miden-crypto (with unified Felt type)

[greenhat]

This PR tests the updated miden-crypto with the unified Felt type.
See 0xMiden/crypto#771 for more details on why we want this.
The miden-crypto PR is 0xMiden/crypto#819. It swaps the Goldilocks as Felt re-export to the Felt(Goldilocks) wrapper.

[bobbinth]

Since I already merged v0.21 VM into main, we'll need to create a branch from the v0.20.6tag and target this PR for that branch (we could call the branchv0.21.7`).

[greenhat]

Since I already merged v0.21 VM into main, we'll need to create a branch from the v0.20.6tag and target this PR for that branch (we could call the branchv0.21.7`).

I created branch v0.20.7 and will rebase this branch onto it.

[greenhat]

Since I already merged v0.21 VM into main, we'll need to create a branch from the v0.20.6tag and target this PR for that branch (we could call the branchv0.21.7`).

I created branch v0.20.7 and will rebase this branch onto it.

Oh, actually I was targeting main with the v0.21 release and NOT v0.20.x so I'm staying targeting main.

Should I remove the v0.20.7 branch, or do you need it anyway?

[greenhat]

The snapshots in the trace generation tests changed. I updated them in the separate commit - e28bc37.

[greenhat]

Rebased and ready.

[greenhat]

The snapshots in the trace generation tests changed. I updated them in the separate commit - e28bc37.

I'm not sure why they have changed thought. The exact change is hard to pinpoint since the diff is hard to read (some changes in the 700 KB one liner).

[plafer]

Upon further investigation, changes in the debug output of the AdviceProvider (part of the ExecutionTrace printed here) are what's causing the diff here. Specifically, Word ordering was changed, which means that Words in the MerkleStore are debug-printed in a different order.

[bobbinth]

Looks good! Thank you!