"Empty" DstData raw bank in calibration/lumi lines rather than no DstData raw bank - wasting bandwidth

When running bandwidth tests over real data and including now the NoBias, Lumi and Hlt2Calib streams in lhcb-datapkg/PRConfig!420 (merged) I came across errors/warnings due to trying to decode empty DstData banks.

@graven and @sesen helped me to understand the issue, and it seems there are some lines in these streams that write "empty" DstData banks, which are in fact not empty but rather 8 bytes plus a generic 8 byte RawBank header, so this is at least 16 bytes / lumi event extra. Better to just not write a bank at all.

The characteristic warning we'll get for this (after LHCb!4565 (merged) is merged) when trying to read back the files will be:

HltPackedBuffer...WARNING DstData raw bank seems to only contain a header, no other data 
HltPackedBuffer...WARNING DstData raw bank has a zero encoding key, and it is not explicitly specified for decoding -- make sure that this really what you want
UnpackDstDataBank WARNING Empty Buffers -- presumably DstData bank has no payload

Doing a bit more digging, it seems that all the calibration.py lines may be causing this to occur in the hlt2calib stream, whilst in the lumi stream the Hlt2LumiDefaultRawBanks and Hlt2LumiCalibration seem to also have this issue (the warning goes away when I comment them out).

This is potentially wasted bandwidth, particularly in the lumi stream. cc @mfontana as I believe these calib lines are under the jurisdiction of RTA Wp4.