There is a requirement to fill gaps. This came up in the telecon on 1 June 2017, see also e-mails around that date. This is related to #4 and possibly #120.
Some users may simply plot the BT array, without looking at the associated lat/lons or scanline numbers. In this case, the gap will not be visible except as a discontinuity in the BTs
Some users may use a filter like in image processing. If this is directly applied to a non-gap-filled array, they will filter together data that shouldn't.
This means that calibration lines also count as gaps, so those also need to be filled — not only missing data in the L1B!
There is a requirement to fill gaps. This came up in the telecon on 1 June 2017, see also e-mails around that date. This is related to #4 and possibly #120.