Unpacker update

[yulunmiao]

This pull request aims at solving the issue Parallelize unpacking per FED and Monitor CM for ZS runs

Parallelizing unpacking per FED is applied using tbb::parallel_for in commit ae4bf46, the affect on timing is still needed to be checked

CM for runs with ZS is saved in a 12 by 2 matrix cm, the element cm(i,j) stands for j-th common mode in i-th eRx subpacket, N.B. that CM in not enabled eRx subpacket may contain garbage. This is in commit 1d27615

[pfs]

hi @yulunmiao sorry this went unnoticed.
thanks a lot. Two requests:

• Can we simplify and store only the sum? <uint16_t, 12, 2 > -> <uint16_t, 12, 1>
• can the we change HGCalECONDPacketInfoSoALayout::location to be always the index of the ECON-D header independently of an exception? We needed for an additional check downstream (in the DQM of the ECON-D CRC)

[yulunmiao]

Hi @pfs,

Thank you for your comments.

I think for CM, there's slightly more information to store separately than store only the sum, and it's reasonable easy to get the sum, so I would suggest keep it the way it is now.

The situation when the location is set not to ECON-D header is

1. Error found in S-Link/Capture Block header
2. Error found in S-Link trailer
   For the first scenario, we assume the package is corrupted, so we don't know where the header we should assign to
   For the second scenario, I am assigning the first ECON-D location to be 0, we can delete this line here to change it
   

   cmssw/EventFilter/HGCalRawToDigi/src/HGCalUnpacker.cc

   Line 343 in 1d27615

   econdPacketInfo.view()[ECONDdenseIdx].location() = 0;

   
   Let me know what I should do.