fix: reduce string comparisons in ADC decoding, cleanup

[baltzell]

Note, this assumes unique crate/slot/channel in all /daq/fadc CCDB tables. Now, the first match wins. Previously the last match won. For run 5038, the RF/LTCC::adc banks have a resulting difference, which will be addressed by fixing CCDB tables to be unique.

Note, this was a component of #1051.

[baltzell]

Summary of hipo-diff -s 0,1,2 on the first 10k events from gold runs for PRs #1064, #1065, and #1066 combined, relative to the development branch:

Run Period     , Gold Run , Redecoderingdiff
rga_spring2018 , 4013     , RF/LTCC::adc
rga_fall2018   , 5038     , RF/LTCC::adc
rgk_2018       , 5700     , RF/LTCC::adc
rgb_spring2019 , 6302     , RF/BAND/LTCC::adc
rga_spring2019 , 6666     , RF/LTCC::adc
rgb_fall2019   , 11567    , RF/LTCC::adc
rgf_2020       , 12933    , RF/LTCC::adc
rgm_2021       , 15833    , RF/BAND::adc
rgc_spring2022 , 16600    , RF/LTCC::adc
rgc_fall2022   , 17407    , RF/LTCC::adc
rgc_spring2023 , 17800    , RF/LTCC::adc and RAW::tdc
rgd_2023       , 18779    , BAND/RF/LTCC::adc
rgk_2024       , 19877    , BAND/RF/LTCC::adc
rge_2024       , 20522    , BAND/RF/LTCC::adc and RAW::tdc
rgl_2025       , 23050    , RF/BAND::adc

[baltzell]

And when CCDB_CONNECTION=sqlite:////work/clas12/users/devita/latest.sqlite:

Run Period     , Gold Run , Redecoderingdiff
rga_spring2018 , 4013     , RF/LTCC::adc
rga_fall2018   , 5038     , RF/LTCC::adc
rgk_2018       , 5700     , RF/LTCC::adc
rgb_spring2019 , 6302     , RF/LTCC::adc
rga_spring2019 , 6666     , RF/LTCC::adc
rgb_fall2019   , 11567    , RF/LTCC::adc
rgf_2020       , 12933    , RF/LTCC::adc
rgm_2021       , 15833    , RF::adc
rgc_spring2022 , 16600    , RF/LTCC::adc
rgc_fall2022   , 17407    , RF/LTCC::adc
rgc_spring2023 , 17800    , RF/LTCC::adc and RAW::tdc
rgd_2023       , 18779    , RF/LTCC::adc
rgk_2024       , 19877    , RF/LTCC::adc
rge_2024       , 20522    , RF/LTCC::adc and RAW::tdc
rgl_2025       , 23050    , RF::adc

[baltzell]

The LTCC/RF discrepancy is due to /daq/fadc/rf and /daq/fadc/ltcc CCDB tables, crate/slot/channel=19/18/15:

clara2501> ccdb dump /daq/fadc/ltcc | grep -E '19[[:space:]]+18[[:space:]]+15'
  19           18           15           79.6         3            15           50           1280            400
clara2501> ccdb dump /daq/fadc/rf | grep -E '19[[:space:]]+18[[:space:]]+15'
  19           18           15           200          3            15           500          0               400

[baltzell]

And the RAW::tdc discrepancy looks like this:

* NODE * group =  20000, item =  12, type = 11, size =       63
         crate :         20        60        64        64        64        64        64
          slot :         20        16        16        16        16        16        16
       channel :         15        20         8         8         8         8         8
         order :         83       -16       -46        52      -106       -13        84
           TDC :        114         0         0         0         0         0         0

* NODE * group =  20000, item =  12, type = 11, size =       63
         crate :         60        20        64        64        64        64        64
          slot :         16        20        16        16        16        16        16
       channel :         20        15         8         8         8         8         8
         order :        -16        83       -46        52      -106       -13        84
           TDC :        114         0         0         0         0         0         0

If it ends up in RAW::tdc/adc, then it's not found in CCDB. So, this one must originate from #1065.

[baltzell]

The RF::adc-only discrepancy for RG-M looks like this:

Bank.show RF::adc
* NODE * group =  21700, item =  11, type = 11, size =       30
        sector :          1         1
         layer :          1         1
     component :          3         4
         order :          0         0
           ADC :       9479       261
          time :    95.1250  132.5000
           ped :        971      2674

* NODE * group =  21700, item =  11, type = 11, size =       30
        sector :          1         1
         layer :          1         1
     component :          3         4
         order :          0         0
           ADC :      10371     34556
          time :     0.0000    0.0000
           ped :        200       200

Maybe/hopefully this will be explained by picking up the pedestal/threshold form /daq/fadc/ltcc in the former case.

[baltzell]

Bank.show LTCC::adc
* NODE * group =  21600, item =  11, type = 11, size =       15
        sector :          5
         layer :          1
     component :          1
         order :          0
           ADC :         96
          time :   118.5625
           ped :        199

* NODE * group =  21600, item =  11, type = 11, size =       15
        sector :          5
         layer :          1
     component :          1
         order :          0
           ADC :          0
          time :     0.0000
           ped :        199

[baltzell]

Another overlap in /daq/fadc/ec and /daq/fadc/ltcc for slot 18, channels 12-15, crates 1, 7, 13, 19, 25, 31 (one for each sector). As /daq/tt/ec doesn't contain these overlapping channels, they should probably be disabled in /daq/fadc/ec by setting one of their indices to bogus.

[baltzell]

After disabling channels 12-15 on all slot 8s in /daq/fadc/ec, the only remaining difference is RF::adc (RF2?), e.g.

* NODE * group =  21700, item =  11, type = 11, size =       30
        sector :          1         1
         layer :          1         1
     component :          3         4
         order :          0         0
           ADC :      24706     44899
          time :     0.0000   31.4375
           ped :        200       200
* NODE * group =  21700, item =  11, type = 11, size =       30
        sector :          1         1
         layer :          1         1
     component :          3         4
         order :          0         0
           ADC :      24706      6378
          time :     0.0000    0.0000
           ped :        200      1754

[baltzell]

/daq/fadc/ftof is sharing with /daq/fadc/rf for crate/slot/channel 23/16/15

[baltzell]

So, after fixing /daq/fadc/ftof and /daq/fadc/ec for crate/slot/channel overlaps, and /daq/tt/band in #1064, and also the RAW::adc/tdc.order software fix in #1084 ...

The first 10k events from each "gold" run's .00001 file are identical between the main branch and redecodering2 (which is the sum of pull requests #1064, #1065, #1066):

Run Period     , Gold Run , `hipo-diff -s 0,1,2`
rga_spring2018 , 4013     , Ok
rga_fall2018   , 5038     , Ok
rgk_2018       , 5700     , Ok
rgb_spring2019 , 6302     , Ok
rga_spring2019 , 6666     , Ok
rgb_fall2019   , 11567    , Ok
rgf_2020       , 12933    , Ok
rgm_2021       , 15833    , Ok
rgc_spring2022 , 16600    , Ok
rgc_fall2022   , 17407    , Ok
rgc_spring2023 , 17800    , Ok
rgd_2023       , 18779    , Ok
rgk_2024       , 19877    , Ok
rge_2024       , 20522    , Ok
rgl_2025       , 23050    , Ok