Fix bug causing problems with joining heterodyned TimeSeries

When joining all the heterodyned time series chunks the join() method of the TimeSeriesList would sometimes fail due to the first time series having an inconsistent time increment dx. I think this was caused by my requirement to delete xindex from each time series and the way that the initial copying of the first time series is done by the join() method. This fix resets the t0 and sample_rate values of each time series after the x indices have been deleted, which appears to fix the problem.

Another minor thing is that time series are now no longer added to the list if they are empty.