Changes to process s2 swath that has more than one granule per tile ID

[CathyJinJNCC]

It turns out a sentinel 2 swath can have two granules for the same tile ID. When input into ARCSI, only one set of ARD files comes back out (probably overwrites the same filenames). EO Team will talk to Pete Bunting about this, but we may need a workaround.

Example swath:
S2B_MSIL1C_20171027T112139_N0206_R037_T30UWF_20171027T133120
S2B_MSIL1C_20171027T112139_N0206_R037_T30UWG_20171027T133120
S2B_MSIL1C_20171027T112139_N0206_R037_T30UWB_20171027T145835
S2B_MSIL1C_20171027T112139_N0206_R037_T30UWC_20171027T145835
S2B_MSIL1C_20171027T112139_N0206_R037_T30UWD_20171027T145835
S2B_MSIL1C_20171027T112139_N0206_R037_T30UWE_20171027T145835
S2B_MSIL1C_20171027T112139_N0206_R037_T30UWF_20171027T145835

[CathyJinJNCC]

Dan suspects it may have been an error with the satellite (so potentially satellite specific?). He's confirmed that processing the duplicate granules separately (e.g. one by itself and the other with the rest of the swath) and then mosaicing the outputs gives a suitable result so we may potentially have to work this into the luigi workflow.

It's worth spending a bit of time checking how often this issue comes up - search for any granules with filenames that are duplicates except for the time part of the timestamp, e.g. S2B_MSIL1C_20171027T*_N0206_R037_T30UWF_20171027T* If more than one result comes up it probably falls under this scenario.