More robust handling of input IDS

Add some safeguards to handle cases when all the time slices are not filled properly, for example when reading experimental data where some profiles can happen to be NaNs. This issue is applicable to all the input IDSs loading.

Start by locating the input IDS loading code and trace how incomplete time slices and NaN profiles from experimental data are handled. Check the behavior across all input IDS loaders; done means safeguards consistently handle unfilled slices without failing.

Written by the indexing model from the issue text.