Check the terminal period of the transition path against a separate resource-constraint tolerance

[arihantlodha-cmd]

What this does

Closes #1210. RC_TPI was a single scalar compared against the resource-constraint error at every period with np.any, so the terminal period could not be exempted without waiving the check across the whole path. The terminal period of a truncated transition is not a true steady state, so its RC error is routinely larger than the interior periods, and the only workaround was to raise RC_TPI globally, which silently allows mid-path violations of the same size (for example OG-UK sets RC_TPI = 0.2 to route around the terminal spike, which then lets a 0.2 interior violation pass).

Changes

• Add an RC_TPI_terminal parameter.
• Check the interior periods (RC_error[:-1]) against RC_TPI and the terminal period (RC_error[-1]) against RC_TPI_terminal.
• _rc_error_message now names whichever tolerance the worst error breaches.

Behavior and compatibility

RC_TPI_terminal defaults to RC_TPI (1e-4), so default behavior is unchanged. A user who needs to exempt the truncation boundary can now loosen RC_TPI_terminal alone and keep the tight interior check, which restores the interior diagnostic that a global loosening removes.

Testing

• A full default-T baseline transition solves with the new check in place.
• tests/test_parameters.py passes (16 passed).
• Added test_rc_error_message_names_terminal_tolerance: the message names RC_TPI_terminal for a terminal-period spike and RC_TPI for an interior one. The existing test_rc_error_message still passes (the helper's third argument is optional).

[codecov-commenter]

Codecov Report

❌ Patch coverage is 54.54545% with 5 lines in your changes missing coverage. Please review.
✅ Project coverage is 75.16%. Comparing base (9baa4b2) to head (6ed370a).

Files with missing lines	Patch %	Lines
ogcore/TPI.py	54.54%	5 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1225      +/-   ##
==========================================
- Coverage   75.17%   75.16%   -0.02%     
==========================================
  Files          24       24              
  Lines        6031     6039       +8     
==========================================
+ Hits         4534     4539       +5     
- Misses       1497     1500       +3     

Flag	Coverage Δ
unittests	75.16% <54.54%> (-0.02%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
ogcore/TPI.py	52.98% <54.54%> (+0.13%)	⬆️

🚀 New features to boost your workflow:

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

[jdebacker]

@arihantlodha-cmd What's the use case for doing this? Is this for some kind of test run? I won't want to make it easier for users to think they've found an equilibrium when they have not as the RC error often points to some model misspecification or a calibration issue.