Deal correctly with missing outputs

Right now if an output value is missing, 0 is assumed and plotted instead.

To demonstrate it, the dataset rt50_missing.tar.gz is just the first two timesteps of the other rt50.tar.gz data where I removed the output values in lines 1, 2, 35, 52, 54, 90, 91, 92. If you select for example "Ca" after loading these data and fix time to 1, the first 2 dots on the left (x=1 and x=2) should not be plotted since they are missing, not 0.