Add uncat function for MultiModalSITS

It implements an uncat_multimodal_sits sits for multimodal sits made of monomodal sits (e.g. from S1 and S2 sits).

For now it just checks if the extracted multimodal sits can be converted to monomodal sits by looking at the mask values.

• Update the constructor of MultiModalSITS to add pack shape information (as used by pack/unpack)
• Update cat_multimodal_sits to propagate/update the ps parameter
• Create uncat_multimodal_sits that splits the sits, check for monomodal conditions and create list of monomodal sits
• Add test in test_time_series.py accordingly