Add worker experience to initial regret update

[kpeluso]

Purpose of Changes and their Description

1. Improve implementation of merit-based sortition -- Calculate percentile with instantaneous scores instead of EMA scores => Faster cycling through actors
2. Resolve runaway regrets
   1. Create migration to clear off regrets and initial regrets
   2. Apply formula developed by @jmdkastro here

      denominator = std(regrets[i-1, :]) + epsilon
      offset = cnorm - 8.25 / pnorm
      dummy_regret[i] = np.percentile(regrets[i-1, :], 25.) + offset * denominator

      which ensures the topic initial regret reaches an appropriate equilibrium.

3. Overwrite topic.AllowNegative to false in all topics -- was wrongly set as true somehow

Link(s) to Ticket(s) or Issue(s) resolved by this PR

• https://linear.app/alloralabs/issue/RES-381/investigate-initial-regret-issue-engineering#comment-2ec7fb4a
• https://linear.app/alloralabs/issue/PROTO-2455/add-experience-to-implement-updated-solution-for-sustain-the-regrets

Are these changes tested and documented?

• [x] If tested, please describe how. If not, why tests are not needed.
• [ ] If documented, please describe where. If not, describe why docs are not needed.
• [x] Added to Unreleased section of CHANGELOG.md?

[github-actions[bot]]

The latest Buf updates on your PR. Results from workflow Buf Linter / buf (pull_request).

Build	Format	Lint	Breaking	Updated (UTC)
✅ passed	✅ passed	✅ passed	⏩ skipped	Oct 14, 2024, 5:56 PM

[kpeluso]

overall new changes lgtm

[spooktheducks]

Not tolerating this kind of response to extremely thorough PR feedback anymore.