Better error handling of interpolate()

[danielhuppmann]

Per issue #806 by @gorkemgungormetu, there are several possible improvements to the interpolate() method if some but not all timeseries have values before and after time (to be interpolated).

1. Clearly list the (head of the) timeseries where data is missing instead of the current generic error message
2. Add an option to ignore_missing (or similar), writing only a warning to the log instead of raising an error