Dud raster scans at low elevation

[ludwigschwardt]

This is a observation + reduction bug... See 1346728350.h5 for an example.

Symptoms: During observation, the source has set below 2.5 degrees elevation but the point_source_scan script keeps going. Each raster scan contains a single dump before the source is skipped (probably because of the new raster label). It takes about a minute before the script is stopped because all sources are down, which is a bit weird. The analyse_point_source_scan script crashes because you cannot fit anything to a single dump.

Fix: Make sure that observation planner correctly terminates script when sources are down. Make reduction script robust against crashes during fitting.