Set include_bias correctly in docs

[kbattocchi]

This makes two changes:

• Changes examples in our documentation that were using a featurizer that includes a bias term while also fitting an intercept; with an OLS final model inference is invalid because they are linearly dependent. This correctly generates warnings if you run the code, but it doesn't cause errors so we never noticed, I've updated the DML docs so that the only remaining warnings are due to convergence issues, not usage related.
• Fixes a small issue with taking the Jacobian of a PolynomialFeatures object, which would incorrectly return nans instead of 0 when called with an argument of 0 (because we were effectively treating a term like y as x^0 y^1, and generating a derivative with respect to x of 0 x^(-1) y^1, which resulted in nan when evaluated at x=0 but 0 as expected when evaluated at any other value of x)

[Copilot]

Pull request overview

This PR addresses two important issues: correcting documentation examples to avoid linear dependence between bias terms and intercepts in OLS models, and fixing a bug in Jacobian calculation for PolynomialFeatures that produced NaN values when evaluated at zero.

• Fixed Jacobian calculation to avoid NaN when X contains zeros by only decrementing positive powers
• Updated documentation examples to set include_bias=False for featurizers when fitting models with intercepts
• Added test coverage for the zero-value case in Jacobian calculation

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

File	Description
econml/utilities.py	Fixed Jacobian calculation in _TransformerWrapper.jac() to avoid NaN by only decrementing positive polynomial powers
econml/iv/sieve/_tsls.py	Applied same Jacobian fix in DPolynomialFeatures.transform() for consistency
econml/tests/test_treatment_featurization.py	Updated test to use include_bias=True and added zero-value test case for NaN fix verification
doc/spec/estimation/dml.rst	Corrected include_bias parameter in examples and updated test setup, but introduced generator expression bugs

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

You can also share your feedback on Copilot code review for a chance to win a $100 gift card. Take the survey.

[fverac]

Looks good