Fixing a bug with success

[jmlarson1]

Fix a bug in the manifold sampling main loop where the successful flag could persist from a previous iteration when termination occurred at the budget nf_max. Previously, if the algorithm exited the inner loop due to reaching nf_max, a stale successful=True value could cause the trust-region center (xkin) to be updated to the most recent evaluation, even when that point was not accepted as an improving iterate.

The fix resets successful at the start of each outer iteration and removes an unnecessary reassignment inside the failure logic, ensuring that only genuinely successful iterations update the incumbent solution.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.59%. Comparing base (d46d537) to head (0570d80).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #304   +/-   ##
=======================================
  Coverage   92.59%   92.59%           
=======================================
  Files         106      106           
  Lines        5011     5011           
=======================================
  Hits         4640     4640           
  Misses        371      371           

☔ 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.

[jared321]

Overall, everything looks find with the PR. We've created Issue #305 to record the lesson learned from this bug.

If @mmenickelly is OK with the fixes, then I approve.

[mmenickelly]

Very necessary, good catch.