break out data fetching from uncertainty calc

This modifies methods of DARMUncertainty that take gpstime inputs to instead take a TimeSeriesDict data object of the exact data for which the uncertainty should be calculated. Data fetching is broken out into a utility function, the output of which can be passed directly to the aforementioned methods.

The data object expected is a gwpy TimeSeriesDict of the relevant TDCF channels, which can be retrieved via the DARMUncertainty.tdcf_channel_list() method:

channels = darm_uncertainty.tdcf_channel_list()
data = gwpy_get_data(channels, gps_start, gps_end)
darm_uncertaionty.compute_response_uncertainty(data, ...)

This simplifies the interface and removes a hard dependency on gwpy in the core model processing.

also:

removed parsing of 'tdcf-data' uncertainty config variable removed redundant calculation from for loop