Fix Stage-1 TC decoding for ECON-T ID 10

[jniedzie]

PR description:

This PR is proposed to fix Stage-1 trigger-cell decoding for P5 data. The unpacker previously assumed that ECON-T ID 10 always represented the second SiPM tile and unconditionally applied the tile offset. In P5 data, this ID can belong to a silicon module, causing incorrect and potentially out-of-bounds reads.

The offset is now applied only when the mapped module is actually a SiPM tile, with an additional bounds check. The Stage-1 buffer is also value-initialized to ensure deterministic behavior when TX words are unavailable.

PR validation:

The PR is a consequence of HGCAL DQM developments, where the old code was replaced by factorized DQM Client plugins. The command used to test the original code is the following:

  cmsRun -j FrameworkJobReport_RAW2DIGI.xml \
    $CMSSW_BASE/src/HGCalCommissioning/Configuration/test/step_RAW2NANODQM.py \
    run=115509 lumi=1 era=P5/v4 \
    daqSourceMode=DTH sourceIdentifier=source \
    buBaseDirsNumStreams=3 buBaseDirsStreamIDs=2100,2101,2102 \
    overrideRangeLS=1,1 \
    files=/eos/cms/store/group/dpg_hgcal/tb_hgcal/2025/DTHP2/run115509 \
    inputTrigFiles=None \
    yamls="{}" \
    output=NANO.root maxEvents=200 \
    skipRecHits=1 \
    enableTPGunpacker=1 \
    secondaryOutput=RAW2RECO.root skipLC=True