[20241209] Merge develop into master

[dannywillems]

master commit: cdbc90c275dba3b70aff5fc2ad5a29ebee10c2e5
developer commit: 872c8f2d1cae6bc0e46d1cf79655eb4592792b52

The code related to Keccak constraints in Kimchi has been removed as it is not used in Berkeley/Kimchi.

Commands:

git checkout master
git checkout -b dw/merge-back-develop-in-master
git merge develop
git rm -rf circuit-construction/
git merge --continue
git push origin dw/merge-back-develop-in-master

In addition to that, I added some commits to make Mina happy.

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 18 lines in your changes missing coverage. Please review.

Project coverage is 76.78%. Comparing base (2136d6c) to head (0c64141).
Report is 2 commits behind head on master.

Files with missing lines	Patch %	Lines
kimchi/src/circuits/expr.rs	0.00%	18 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2868      +/-   ##
==========================================
+ Coverage   76.61%   76.78%   +0.17%     
==========================================
  Files         255      253       -2     
  Lines       60855    60614     -241     
==========================================
- Hits        46622    46543      -79     
+ Misses      14233    14071     -162     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[querolita]

After checking with Danny, I confirm that the Keccak parts that are being removed in this PR are not in use right now. Those constraints and gadgets were a first version of the ones that were then introduced for the o1vm project, and they were only being used inside an old-ish draft PR to check correctness of constraints and witness generation (which required a generic number of witness columns, as well as a full integration of that circuit gate in the proof system, also closed PRs). Other than that, we must not remove the keccak witness generation code inside kimchi, because that's the one being used in o1vm. Perhaps now it makes sense to move that inside the o1vm module.

I also checked that the changes I made to the bound on the zk rows for chunking has been correctly updated in this merge.

[volhovm]

This seems about right -- I understand the only last 3 commits are "on top of master for compatibility with mina". They seem to be very benign.