fixed issue parsing numpy array of times when writing irregular ts

[klarrieu]

When writing an irregular time series, there is a check to make sure that times exists as is either an instance of list or tuple. However, while numpy arrays work fine for regular time series, it fails for irregular time series. In this PR I amended the problematic code to also accept numpy arrays when writing irregular time series. This check fails in its current form when given numpy arrays both because if not times is ambiguous for an array and isinstance(times, (list, tuple)) evaluates to False, so I changed if not times to if times is None (less ambiguous) and I allowed times to be an instance of a numpy.ndarray for irregular time series as well.

[gyanz]

@klarrieu Can you modify the code to check if the length of the time container (array, list, or tuple) is not zero? len function should work for all types of containers.

[klarrieu]

Thanks for taking a look at this! I agree that is a good call. I adjusted the code for this PR accordingly.