Fix SQRT fn for secp256r1 after accidental merge about halo2curves

[rrrliu]

All good - I'm out today but have a lot of time tomorrow to fix!

[huitseeker]

@CPerezz since we have both a merge queue and #54, we may want to close this.

[huitseeker]

This should be fixed in #54, but for the purpose of avoiding this sort of issue occurring again, perhaps this repo could use a GH merge queue?

TL;DR: this alleviates merge mistakes and time-of-fork vs. time-of-merge CI discrepancies by running CI as part of the merge process.

https://github.blog/changelog/2023-02-08-pull-request-merge-queue-public-beta/
https://docs.github.com/en/repositories/configuring-branches-and-merges-in-your-repository/configuring-pull-request-merges/managing-a-merge-queue#triggering-merge-group-checks-with-github-actions

The CI changes are ~3 lines which I can totally tack on #56
(there are settings changes in the branch protection rules, which an admin of the repo would have to do ofc)