Bug with SHAP library v.0.41

This bug was present in some SHAP plots (such as in the bar plots):

shap.plots.bar(shap_values_game)

---------------------------------------------------------------------------
UnboundLocalError                         Traceback (most recent call last)
Cell In[47], line 1
----> 1 shap.plots.bar(shap_values_game)

File ~/anaconda3/envs/shap_env2/lib/python3.9/site-packages/shap/plots/_bar.py:54, in bar(shap_values, max_display, order, clustering, clustering_cutoff, merge_cohorts, show_data, show)
     51     assert isinstance(shap_values, dict), "You must pass an Explanation object, Cohorts object, or dictionary to bar plot!"
     53 # unpack our list of Explanation objects we need to plot
---> 54 cohort_labels = list(cohorts.keys())
     55 cohort_exps = list(cohorts.values())
     56 for i in range(len(cohort_exps)):

UnboundLocalError: local variable 'cohorts' referenced before assignment

This bug was fixed in the 0.42 release (https://github.com/shap/shap/releases/tag/v0.42.0); however, econml has SHAP<=0.41 as a requirement, so these issues will persist until the requirements for the SHAP version are updated.

Review EconML's SHAP<=0.41 requirement and the SHAP v0.42.0 release linked in the issue, then locate where the dependency constraint is defined. Update the supported SHAP version and verify that the affected SHAP plots no longer encounter the reported error.