Replacing No-data channels with dummy dc timeseries

A recent addition to the pipeline enforced the data from channels that are zero (or don't have data for the analysis segments) to be replaced with unity at all the samples, as a dc time-series will not have any correlation with the DARM. This was done at the level of reading the frames using the .read() function within the timeseries.py.

• Check if this is indeed working or is the right way to do it.
• If the above doesn't work, think of replacing the channels that don't;t have data so that at the time of inference, they will not be read at all.

Start in timeseries.py at the .read() path that replaces zero or missing channels with unity. Check whether this behavior works for analysis segments and whether it is appropriate for inference; if not, evaluate excluding channels without data. Done means the chosen handling is verified and documented by the relevant checks.