Robust handling of duplicate and overlapping model parameters

[mjrutala]

At present, overlapping time spans in the read_SWModel() routines are only partially handled, and overlapping time spans with identical datetime indices, or repeated datetime indices within the same SWModel input file, are not handled at all.

While repeated times should not be common, overlapping indices are, as in models which span a fixed length of time (i.e., a Carrington rotation + a few days) which need to be stitched together to create a longer timeseries. Both these issues, should, however, be able to be handled the same way.

Routines need to be added to all read_SWModel() model readers which:

1. Detect repeated datetime indices in an input file
2. Detect repeated datetime indices in the combined output
3. Detect overlapping (but not necessary identical) datetime spans in the combined ouput
4. Average overlapping model parameters together, reindexing in the case of non-identical overlapping datetime spans