Allow development to accept an age in months relative to origin

[Abhayindia]

Closes #1094 (the enhancement route agreed with @henrydingliu in that thread).

_to_datetime tries %Y%m, %Y, %Y-%m-%d and then falls through to a bare pd.to_datetime, which for a numeric column silently reads the values as nanoseconds since epoch. That collapsed age-lag columns like [12, 24, 36] to 1970-01-01, and the old guard just checked whether the result landed in 1970 - so a triangle whose development really is the year 1970 got the same error.

This replaces that with a check at the point of parsing: if none of the explicit formats matched and the column is numeric, it's an age in months. _init_development then resolves each value to the valuation date that many months after the start of that row's origin period. Ages are measured from the origin period start (using the constructor's own origin_grain, so mid-period origin dates and fiscal-year anchors line up), not the literal recorded origin date. Bare years still match %Y and are untouched. Non-numeric unparseable input still raises, with the same message as before. Semiannual grains only support the calendar Jan/Jul anchor; a non-calendar one raises a clear error rather than guessing.

Verified:

$ python -m pytest chainladder/core/tests/test_triangle.py -q
312 passed, 1 skipped in 100.59s

Full suite was 1091 passed / 8 skipped when I posted the branch on the issue.

[github-actions]

Pyright Type Completeness

View the full pyright --verifytypes output for this commit

Project (full chainladder package, at this PR's head): 15.3% of exported symbols fully typed (211 / 1379)

	Known	Ambiguous	Unknown	Total
Project (head)	211	111	1057	1379

Other symbols referenced but not exported by chainladder: 13

	Known	Ambiguous	Unknown	Total
Other (head)	3	1	9	13

Symbols without documentation:

• Functions without docstring: 325
• Functions without default param: 0
• Classes without docstring: 10

Patch (exported symbols added or changed by this PR): 100.0% fully typed (3 / 3); 1 no longer exported

	Known	Ambiguous	Unknown	Total
Patch	3	0	0	3

Patch symbol details

Symbol	Status	Change
chainladder.core.tests.test_triangle.test_set_development_age_in_months	✅ known	new
chainladder.core.tests.test_triangle.test_set_development_age_respects_mid_period_origin	✅ known	new
chainladder.core.tests.test_triangle.test_set_development_bare_years_unaffected_by_age_support	✅ known	new
chainladder.core.tests.test_triangle.test_set_development_age_instead_of_date_raises	—	no longer exported (was ✅ known)

[henrydingliu]

just unblocked tests. expecting ruff and codecov fails.

[codecov]

Codecov Report

❌ Patch coverage is 62.50000% with 6 lines in your changes missing coverage. Please review.
✅ Project coverage is 91.61%. Comparing base (16460a8) to head (4a0ad6d).
⚠️ Report is 3 commits behind head on main.

Files with missing lines	Patch %	Lines
chainladder/core/base.py	62.50%	4 Missing and 2 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1229      +/-   ##
==========================================
- Coverage   91.70%   91.61%   -0.10%     
==========================================
  Files          93       93              
  Lines        5435     5448      +13     
  Branches      699      703       +4     
==========================================
+ Hits         4984     4991       +7     
- Misses        327      331       +4     
- Partials      124      126       +2     

Flag	Coverage Δ
unittests	91.61% <62.50%> (-0.10%)	⬇️

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.

[henrydingliu]

moving to draft due to conflicts

[Abhayindia]

@henrydingliu the conflict is genedan's #1088, which added test_set_development_age_instead_of_date_raises - it asserts development-as-age raises ValueError, the opposite of what this PR does. @genedan do you still want the age-in-months feature from #1094, or has the direction shifted to "reject ages with a clear error"? happy to rework either way.

[henrydingliu]

good to deprecate test_set_development_age_instead_of_date_raises

[Abhayindia]

rebased on main - kept the age-in-months tests and dropped test_set_development_age_instead_of_date_raises per your call. ruff check clean; the ruff format fail is the same pre-existing whole-file base.py/test_triangle.py nonconformance.

[henrydingliu]

the ruff format fail is the same pre-existing whole-file base.py/test_triangle.py nonconformance

is it a bug that the per-file ignore didn't work?

codecov still shows uncovered changes

[Abhayindia]

per-file-ignores only apply to ruff check, not ruff format - the formatter has no per-file exclusion, so any PR that touches base.py trips the whole-file format check. base.py needs ~200 lines of reformatting, which really belongs in its own cleanup PR (same situation #1216 describes). I can run ruff format on test_triangle.py here (~60 lines) if that helps, but base.py I'd rather leave.

codecov: the uncovered lines are the semiannual-origin (2Q) branch of the age path - it only triggers for a non-default semiannual anchor and is awkward to exercise in a unit test. Happy to add one if you want it covered.

[henrydingliu]

understood on ruff format.

it only triggers for a non-default semiannual anchor and is awkward to exercise in a unit test.

does the new prism triangles help?