Fix timestamp: subtract packet time instead of full scan time

[MichaelGrupp]

The previous timestamp patch in PR #104 subtracted the full scan time from the last point time of the first packet to determine the scan timestamp.

It has to be the time for a single packet instead. Otherwise, the timestamp is too far in the past.

This is especially noticeable when the scan frequency is in a lower range like 10Hz, i.e. the scan timestamp would be 100ms off compared to other sensor modalities.