argument `X` in `model` function is not always a pandas Dataframe

[folivetti]

In the CONTRIBUTING.md file it is mentioned that the function model is going to be called by passing the estimator and a pandas DataFrame:

https://github.com/cavalab/srbench/blob/master/CONTRIBUTING.md###-returning-a-sympy-compatible-model-string

But it seems that sometimes the second argument is a numpy array. We should double check that before release.
A workaround is this implementation from Bingo https://github.com/cavalab/srbench/blob/master/algorithms/bingo/regressor.py