CubicInterp on lower bound returns np.nan

[alanlujan91]

Describe the bug
Evaluating CubicInterp on x_list.min() produces np.nan.

To Reproduce
See example notebook https://gist.github.com/alanlujan91/ee468fb3e83c411eaccbfb22d4943a69

Expected behavior
The problem is solved when lower_extrap=True, but evaluating at x_list.min() is not lowe extrapolation, should be evaluated by default. In some instances we might want lower_extrap=False and interp(x_list.min()) = y_list.min().

Desktop (please complete the following information):

• OS: Windows
• Browser: Chrome
• Econ-ARK and Python version: 0.10.8 and python 3.8.8