CausalTree split criterions fix and fit optimization

[alexander-pv]

Proposed changes

Hi, this PR contains several fixes for causal trees issues I have recently found:

1. It turned out that sum_total and sq_sum_total attributes in RegressionCriterion were calculated using weights as multipliers. However, in CausalTree these weights have two values: eps for control, 1 for treatment. So, total outcome values for a tree split were incorrect. Link: _criterion.pyx

2. There was incorrect calculation of weighted_n_node_samples attribute in RegressionCriterion. It is important to get the right impurity values and feature importance. Links: _criterion.pyx, _tree.pyx

3. There is a more efficient way of calculating node and children impurity for each split. CausalTree fit() measurements clearly show this:

Additional code for time measurements: test.zip
Related issue: #541.

So, I added NodeInfo and SplitState structs in criterion.pyx for clarity and updated CausalRegressionCriterion. These steps allowed me to get rid of the loops in CausalMSE and even make this class more concise.

Curiously, the qini score with synthetic data has slightly changed since this update.

Additional updates:

• CausalTreeRegressor fit_predict() method calculates bootstrap-based treatment effect along with bootstrap-based bounds.
• Notebooks with causal trees
• Tests for causal trees
• Minor update in conftest.py to get the desired type of regression synthetic data in any test file.

Types of changes

What types of changes does your code introduce to CausalML?
Put an x in the boxes that apply

• Bugfix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation Update (if none of the other choices apply)

Checklist

Put an x in the boxes that apply. You can also fill these out after creating the PR. If you're unsure about any of them, don't hesitate to ask. We're here to help! This is simply a reminder of what we are going to look for before merging your code.

• I have read the CONTRIBUTING doc
• I have signed the CLA
• Lint and unit tests pass locally with my changes
• I have added tests that prove my fix is effective or that my feature works
• I have added necessary documentation (if appropriate)
• Any dependent changes have been merged and published in downstream modules

Further comments

[jeongyoonlee]

This is great! Thanks for your contribution. I left one comment regarding the use of the bootstrapped mean. Otherwise, LGTM.

Thanks!

[jeongyoonlee]

LGTM. Thanks!