Add default PyMC models when model=None for all experiment classes

[ErikRingen]

Summary

Closes #661.

Experiment docstrings claimed default PyMC models would be used when model was omitted, but BaseExperiment immediately raised ValueError("model not set or passed."). This aligns actual behavior with the documented API.

• Add _default_model_class class attribute to BaseExperiment with fallback logic that instantiates the default when model is None
• Set appropriate defaults on all 9 experiment subclasses: LinearRegression for most, WeightedSumFitter for SyntheticControl, InstrumentalVariableRegression for InstrumentalVariable, PropensityScore for InversePropensityWeighting
• Fix incorrect type hints (BaseExperiment -> PyMCModel) in InstrumentalVariable, InversePropensityWeighting, and RegressionKink
• Remove redundant self.model = model in InstrumentalVariable that would have overwritten the default set by super().__init__()

Test plan

• New parametrized integration test covers all 9 experiment classes with model=None
• Explicit model precedence test confirms user-provided models are not overridden
• Existing test_model_experiment_compatability tests pass (OLS rejection still works)
• Full CI integration test suite passes

Made with Cursor

[cursor]

PR Summary

Medium Risk
Changes default initialization behavior across many experiment entrypoints, which could affect downstream expectations and sampling defaults; coverage is added but runtime/model-configuration differences may still surface in user workflows.

Overview
Experiment subclasses now provide a documented default PyMC model when model=None, by adding BaseExperiment._default_model_class and instantiating it automatically before the existing model-type support checks.

All affected experiments set _default_model_class (mostly LinearRegression, with WeightedSumFitter for SyntheticControl, InstrumentalVariableRegression for InstrumentalVariable, and PropensityScore for InversePropensityWeighting), update docstrings/type hints accordingly, and remove an InstrumentalVariable assignment that would have overwritten the default. Adds a parametrized integration test (test_default_models.py) to ensure defaults are used and that explicitly passed models still take precedence.

^{Written by Cursor Bugbot for commit e38fa9e. This will update automatically on new commits. Configure here.}

[read-the-docs-community]

Documentation build overview

📚 causalpy | 🛠️ Build #31370886 | 📁 Comparing e38fa9e against latest (323e657)

🔍 Preview build

Show files changed (22 files in total): 📝 22 modified | ➕ 0 added | ➖ 0 deleted

File	Status
404.html	📝 modified
api/generated/causalpy.experiments.diff_in_diff.DifferenceInDifferences.html	📝 modified
api/generated/causalpy.experiments.instrumental_variable.InstrumentalVariable.init.html	📝 modified
api/generated/causalpy.experiments.instrumental_variable.InstrumentalVariable.html	📝 modified
api/generated/causalpy.experiments.inverse_propensity_weighting.InversePropensityWeighting.init.html	📝 modified
api/generated/causalpy.experiments.inverse_propensity_weighting.InversePropensityWeighting.html	📝 modified
api/generated/causalpy.experiments.prepostnegd.PrePostNEGD.html	📝 modified
api/generated/causalpy.experiments.regression_discontinuity.RegressionDiscontinuity.html	📝 modified
api/generated/causalpy.experiments.regression_kink.RegressionKink.init.html	📝 modified
api/generated/causalpy.experiments.regression_kink.RegressionKink.html	📝 modified
api/generated/causalpy.experiments.staggered_did.StaggeredDifferenceInDifferences.html	📝 modified
api/generated/causalpy.experiments.synthetic_control.SyntheticControl.html	📝 modified
_modules/causalpy/experiments/base.html	📝 modified
_modules/causalpy/experiments/diff_in_diff.html	📝 modified
_modules/causalpy/experiments/instrumental_variable.html	📝 modified
_modules/causalpy/experiments/interrupted_time_series.html	📝 modified
_modules/causalpy/experiments/inverse_propensity_weighting.html	📝 modified
_modules/causalpy/experiments/prepostnegd.html	📝 modified
_modules/causalpy/experiments/regression_discontinuity.html	📝 modified
_modules/causalpy/experiments/regression_kink.html	📝 modified
_modules/causalpy/experiments/staggered_did.html	📝 modified
_modules/causalpy/experiments/synthetic_control.html	📝 modified

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.40%. Comparing base (323e657) to head (e38fa9e).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #707      +/-   ##
==========================================
+ Coverage   94.36%   94.40%   +0.04%     
==========================================
  Files          44       45       +1     
  Lines        7591     7646      +55     
  Branches      461      463       +2     
==========================================
+ Hits         7163     7218      +55     
  Misses        264      264              
  Partials      164      164              

☔ View full report in Codecov by Sentry.
📢 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.