Fix and enhance utils.model diagnostics part 3

[henrydingliu]

Summary of Changes

fixing bug on latest
adding support for predicted triangles
adding more tests
simplifying some existing tests

Related GitHub Issue(s)

Additional Context for Reviewers

• I passed tests locally for both code (uv run pytest) and documentation changes (uv run jb build docs --builder=custom --custom-builder=doctest)

---

Note

Medium Risk
Changes diagnostic outputs (Latest/IBNR) that downstream users may rely on, though behavior is now aligned with ultimate and latest diagonal; scope is limited to a utility helper plus tests.

Overview
model_diagnostics is updated so Latest and IBNR line up with triangle semantics: loss data is always taken through cum_to_incr() (replacing the prior incr_to_cum() path when not grouping), and IBNR is Ultimate − Latest instead of obj.ibnr_. The helper also accepts a fitted/predicted Triangle, rejects inputs without ultimate_ / ibnr_ / ldf_, and documents Triangle | MethodBase | Pipeline in its signature.

Tests add np.allclose checks that model_diagnostics matches latest_diagonal and ultimate_ on fit and predict across several estimators, a groupby parity test on prism, and a ValueError when passing a raw triangle; a few predict tests are simplified (fixtures, sample-weight expectations, clrd ordering).

^{Reviewed by Cursor Bugbot for commit fd8af0b. Bugbot is set up for automated code reviews on this repo. Configure here.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 89.05%. Comparing base (cdd8371) to head (fd8af0b).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #959      +/-   ##
==========================================
+ Coverage   88.96%   89.05%   +0.09%     
==========================================
  Files          89       89              
  Lines        5100     5072      -28     
  Branches      663      650      -13     
==========================================
- Hits         4537     4517      -20     
+ Misses        417      415       -2     
+ Partials      146      140       -6     

Flag	Coverage Δ
unittests	89.05% <100.00%> (+0.11%)	⬆️

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

☔ 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.