Timeseries to table transformation: Fix misaligned table field values if some frames are missing a label

[domasx2]

Hey!

Found a bug with converting frame labels to fields in time series table transform.

When coverting time series frames to a "table" frame, a field is created for each label and filled with label value from each frame. However if some frame in the middle is missing a particular label, transform would just skipt it and not add an empty value. Thus field values would become misaligned with other fields.

For example if all frames have name label but only some frames have namespace label, in the resulting table namespaces would become misattributed to wrong names. Names coming from frames with missing namespace would get attributed namespace from the next frame that has one.

It's probably best to look at the added test to quickly grok the issue :)

[codeincarnate]

Looks good and the test makes sense to me 😄. Tested out on a dashboard too and saw the error with labels floating up and this definitely seems to fix that 👍