fix: witnesscalc build

[Autoparallel]

To verify this is all working, you can run:

build-circuit circuits/aes-gcm/aes-gcm-fold.circom aes-gcm-fold.bin -l node_modules

to generate the graph file for witness extension.

To see the constraint count for aes-gcm-fold, you can run:

circom circuits/aes-gcm/aes-gcm-fold.circom -l node_modules --r1cs

---

Currently addresses:

• Closes https://github.com/pluto/web-prover/issues/215

---

Need to solve:

• Makes progress towards golf: reduce constraint count of aes-gcm-fold with respect to circom-witnesscalc #93
  Which the following may help with:
• Closes golf: use the lookup into FF inverse method to save constraints #94, addressed with 89da3ee

[Autoparallel]

@0xJepsen @thor314 I will need your help to get this to pass tests and what not.

Also the constraints for aes-gcm-fold are now near 1M. We need to reduce this heavily by either folding more or ideally optimizing. Folding more is fine for now. I don't know where the good folding boundaries are, so I'm leaving that for you two to decide on.

[Autoparallel]

All tests are passing, but I would appreciate reviewers giving this another look to make sure that this is still totally working as expected.

[Autoparallel]

Potential improvement noted here

aes-proof/circuits/aes-gcm/aes-gcm-foldable.circom

Line 105 in f0b1dcc

// TODO: Remove this blocks-to-stream translation by migrating the ghash select

[devloper]

I reviewed this when I updated the tests and confirmed that the changes are working as expected. From my PoV this is ready to go, the primary blocker atm is that it's dependent on the PP PR (the circuit interface changes)

We can probably merge this now, given that upstream repos should be pinned to a specific version of aes-proof.

[thor314]

• aes-gcm-fold is difficult to read as is, could use comments
• replace todos
• many magic numbers in Cipher, consider replacing with named variables for readability
• ff circom could use some comments to explain the longer but more efficient approach

If helpful, my espanso todo replacement snippets:

# espanso replace snippet
  - triggers: ["TODO ", "tdd "]
    replace: "TODO(TK {{date}}): "  
    vars:
      - name: date
        type: date
        params:
          format: "%Y-%m-%d"
  - triggers: ["tdb "]
    replace: "- [ ] TODO(TK {{date}}): "  
    vars:
      - name: date
        type: date
        params:
          format: "%Y-%m-%d"
  - triggers: ["tdc "]
    replace: "// TODO(TK {{date}}): "  
    vars:
      - name: date
        type: date
        params:
          format: "%Y-%m-%d"
  - triggers: ["tdp "]
    replace: "TODO(TK): "  

[thor314]

@Autoparallel, this looks good to go after the comments I made above

[0xJepsen]

nice work on this

[thor314]

@devloper @Autoparallel Would you annotate your todos:

TODO(<initials> <date>): <comment>

e.g.
TODO(TK 2024-10-21): readability: add comments about what this does

a bare TODO is nice for working alone, but it doesn't say much about how old the thing is for the reader; makes me git blame to see all the relevant metadata