Investigate/Fix empty output for first second of VDIF data

[robotopia]

No description provided.

[bwmeyers]

@cplee1 Is investigating. It appears that the first second of data is essentially random or near-zero data, indicating it's related to accessing a buffer before it's been populated correctly. Comparing the PSRFITS and VDIF timestamps indicates that they are matching and correct if the first second is ignored, so it really does look like an extra second of "nothing" is being written for no apparent reason.

[cplee1]

Below is a screenshot of the native-resolution detected timeseries output from the combination of CDMT and dedisp at a DM of 0. It seems that the first second is mostly constant followed by some offset chunks at the edges of the timeseries, presumably from scales/offsets computed at the edge. I'll also add that the last chunk of the timeseries is always offset as well.

[bwmeyers]

See #45 (comment)