Add return_array argument to CoxPHSurvivalAnalysis().predict_survival_function

[rv-apuri]

Is your feature request related to a problem? Please describe.
It is mildly frustrating that the predict_survival_function for CoxPHSurvivalAnalysis() does not take in the return_array argument in the same way as RandomSurvivalForest().predict_survival_function does

Describe the solution you'd like

estimator = CoxPHSurvivalAnalysis().fit(X, y)
survs = estimator.predict_survival_function(X.iloc[:5], return_array=True)

survs should be an ndarray containing the values from the step function

Describe alternatives you've considered
A quick workaround to get the array is below:

estimator = CoxPHSurvivalAnalysis().fit(X, y)
funcs = estimator.predict_survival_function(X.iloc[:5])
survs = np.array([f(f.x) for f in funcs])

References and existing implementations
If you want to propose a new algorithm for inclusion, please include the original reference to the publication that first proposed the algorithm. If you are aware of any existing implementations, e.g. for R, please include a link too.

[sebp]

I agree that this would be quite useful.

[rv-apuri]

Thank you so much!!