RP interpolation & fill value

[zackarno]

          Yup looks good! 

Only thing I noticed- in admins where the minimum yearly maximum is greater than 0 (i.e. there was some amount of flooding every year), the RP of a 0 value is 1.0 years (due to the interpolation being out of range, and using the lower fill_value of 1). However, if the admin has at least one year where there was no flooding, then the RP of a 0 value is 1.038462 (i.e. (len(group) + 1) / len(group)), because it's then inside the bounds of the interpolation.

Anyways of course this is splitting hairs since they both get reclassified as 1-1.5. So makes no difference for the published dataset.

Originally posted by @t-downing in #11 (review)