spline_pdf efficiency issue

[maxwest-uw]

when trying to get scipy.stats (mean, var, std etc.) for a spline_pdf ensemble, the runtime is increased exponentially when compared to every other pdf function.

• discovered this when trying to add a _run_ensemble_funcs test for splines in test_ensemble.py, with a single test runtime jumped from ~30 seconds to 90 seconds
• I was able to narrow it down to the fact that the spline_pdf class is the only qp pdf function that doesn't implement its own ._ppf method. the native scipy way of calculating ppf for splines seems to be very inefficient and gets called multiple times when trying to calculate the mean (due to the fact that scipy has to integrate over the function)
• should be able to make it run faster if we do a similar implementation to the one in interp_pdf._ppf (as suggested by Alex)