GitLab

realising that TGV or Scaler readout can be split between the end of one
buffer and the start of the following, in which case the beginning
of the following buffer is to be ignored until an 'EXT-TS' and its
accompanying friend '0x1000*' have been read.

 * data begins with START_READOUT dummy module and ends with END_READOUT
   dummy module
 * TGV is not expected to be last (or first)
 * we do not require all modules to be present, or in order, just to have
   a known module id, and occur in between START_READOUT and END_READOUT

 * read_event_in_buffer has to be called with the MFM frame revision

* modules with no data are no longer included
* 'EOE' words are not included either

i.e. everything starting with 0xF*: 0xF1, 0xF3, 0xF9, etc.

...now that we understand the structure

and so there is no reason to expect each buffer to begin with the
readout of the first module each time!

this happens especially as the acquisition rate / data flow increases...

 * prints name of detector associated with each hit module/channel

 * consistency check based on order of readout of modules

 * dummy setup with just number of modules no longer allowed
 * the crate map definition must always be given

 * trying to understand problems reading data (finalement this doesn't help)

 * allows to use timestamp in EoE

 * in order to retrieve timestamp for each module (checking)

 * in case EOE with timestamp is used