FIX: TimeSeries.align() samples the x-axis at the least frequent TimeStream's granularity

[jonahjlee]

In time_series.py:

@staticmethod
    def align(*args) -> list:
        start_time = np.max([arg.start.timestamp() for arg in args])
        end_time = np.min([arg.stop.timestamp() for arg in args])
        granularity = np.max([arg.granularity for arg in args])
        new_length = end_time - start_time

        new_x_axis = np.linspace(0, new_length, math.ceil(new_length / granularity))

Since granularity has units of seconds, higher granularity means lower frequency data. The intended behaviour was to sample the most frequent TimeStream to preserve all data, and this could be extended to allow for users to pass in a custom new granularity.