Patch 326: validate run summary channels

[kujaku11]

This PR addresses Issue #326 and adds a validation to the RunSummary to check that the input and output channels are consistent between runs. If the channels are not a logger warning is produced. If the use wants they can drop those runs that are inconsistent. Updated tests.

• Comment out drop_nans in transfer_function_helpers | process_transfer_functions as the nans are taken care of later. This allows for accounting of different channels.

• Added test in xarray_helpers | handle_nan if the merge does not work with join='exact' then try 'join='outer'

• Updated decimation method to use scipy.signal xarray accessors built by mth5.timeseries.scipy_filters

• Tests pass

[kkappler]

@kujaku11 This looks like is catches a useful edge case. As RunSummary is now in mtpy-v2, can you take a look if this can be merged there instead of aurora?