FM sample issue: no smearing in SimChannel, not using DNNROI

[HaiwangYu]

• SimChannel smearing for dune10kt-1x2x6/wcls-sim-drift-simchannel-nf-sp.jsonnet

// >>> ADD THESE TWO LINES <<<
    // Smear the truth SimChannel to reco (SP) resolution so the 2D truth labels
    // register onto the deconvolved reco charge (added in quadrature to the
    // post-drift depo diffusion). Values = dune10kt-1x2x6 SP-filter recipe:
    //   smear_long [raw ticks] = sigma_t/tick, sigma_t = 1/(2*pi*0.12MHz) = 1.326us
    //   smear_tran [pitch] = 1/(2*sqrt(pi)*k): Wire_ind k=0.75 (U,V); Wire_col k=3.0 (W)
    smear_long: 2.6526,
    smear_tran: [0.37612, 0.37612, 0.09403],
  },

• Make use_dnnroi default, use the new 6 channel version.

• experiment to fix:

  • PD: protodunevd, pdhd
  • FD: dune10kt-1x2x6, dune-vd, dune10kt-hd, dune10kt-vd
  • For FD need to confirm with Dom when adding DNN as default?
  • SBND

• Validation: for each experiment

  • configure a standalone workflow with ideal track at 0, 45 and 85 degrees
    • the configuration should use same configuration parameters as the larsoft version other than the input/output part
  • run the sim/simchannel/sp(dnn) chain
  • compare the decon and sim channel waveforms

[HaiwangYu]

Work note:
https://hackmd.io/@HaiwangYu/SJCWWbUQMl

Summary:
https://github.com/HaiwangYu/dunereco/blob/b220feb6a45e88eea0d152043e9b96ccd009864f/dunereco/DUNEWireCell/docs/smear-dnn-campaign.md

Result for dune10kt-1x2x6, w/o smearing:
https://github.com/HaiwangYu/cffm-if/blob/main/smear-dnn/dune10kt-1x2x6/deg00/plots/
https://github.com/HaiwangYu/cffm-if/blob/main/smear-dnn/ablation-1x2x6-nosmear/deg00/plots/