fix: use 0 instead of null values to avoid breaking the plots

[jfrer]

closes #40

[paulpestov]

Not sure about this. Replacing null by 0 would display wrong data in the charts since 0 is not some default value but an actual measurement. We should have a strategy how to deal with nulls in the data. @MareenGeestmann @mweidling

[jfrer]

Not sure about this. Replacing null by 0 would display wrong data in the charts since 0 is not some default value but an actual measurement. We should have a strategy how to deal with nulls in the data. @MareenGeestmann @mweidling

Hmm yes, this could be problematic as the user won't know the difference between values that are actually 0 and missing values. However this is also the way that null values are currently handled in the table view (see e.g. the CER standard deviation for "Bettina von Arnim. Tagebuch. 1835." where the values are null for each workflow but 0 is displayed instead).

Also I think that 0 is the correct value for the standard deviation in these cases (because the CER range has the same start- and end-value), but we can't be sure for every other case of course.

[mweidling]

IIRC we decided that the runs where we get nulls should not be considered in the plots but skipped. I don't know how much effort it is to achieve this, though.