Fixing encoder-decoder bugs in [audit 2026-07-09], Add fix-prompt-applier skill

[odunbar]

Purpose

• Closes [audit 2026-07-09] identified a variety of encoder-decoder issues. #449
• Adding fix-prompt-applier: On construction of the code-review I create fix-prompts Markdowns which are prompts to help fix the bugs found. fix-prompt-applier is designed to carry-out the fixes based on these prompts, and update the review.md with status/progress of resolution

Content

Fixes to Close #2, focused on encoder-decoder and robustness

• is_initialized(...) method checks if DataProcessors have been initialized. If it is, then it throws a warning that the user is using an old processor (allows for reuse of processors, but tells the user in case of mistake)
• some better rank-stability for CanonicalCorrelation processor
• decorrelator randomized norm estimator (which does frobenius norm) swapped to trace estimator for consistency with the small matrix setting. Otherwise the criteria for truncating the matrix changes for the small and large dimension setting
• bugfix an off-by-one error in the stencil for likelihood informed processor
• extend the MCMC methods to have nonzero mean in processed space.
• Various robustness catches (e.g. some singular division guards for RF, and encoders, equality bypassing, correctly using passed rngs)

---

• I have read and checked the items on the review checklist.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 95.38%. Comparing base (dc4dea8) to head (e32eb06).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #450      +/-   ##
==========================================
+ Coverage   95.19%   95.38%   +0.19%     
==========================================
  Files          12       12              
  Lines        2495     2537      +42     
==========================================
+ Hits         2375     2420      +45     
+ Misses        120      117       -3     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.