Decoding speedups

[baltzell]

Summary:

• Less string comparisons and looping
• More hash/index-based access for CCDB tables and banks
• Use new global TranslationTable class, with additional DetectorType field
• Reduce EVIO branch/node transversing

Side Effects:

• Run 5038
  • Row-ordering changed for ECAL::tdc, RAW::vtp, BMT::adc, and RAW::adc
  • CCDB overlap for /daq/fadc/ltcc and /daq/fadc/rf (LTCC::adc and RF::adc)
• Run 18799
  • CCDB overlap for /daq/tt/band and /daq/tt/raster (BAND::adc and RASTER::adc)

Note, this is the completion of #1008 ...

[baltzell]

For 5038, /daq/fadc/ltcc and /daq/fadc/rf CCDB tables have a duplicate crate/slot/channel.

And, hipo-diff also isn't happy with ECAL::tdc, RAW::vtp, BMT::adc, and RAW::adc:(

[baltzell]

RAW::vtp is only different by row ordering (now it's sorted by crate, the first index).

[baltzell]

ECAL::tdc is also only different by row ordering (now sorted by s/l/c/o).

[baltzell]

And BMT::adc and RAW::adc are also only different by row ordering.

[baltzell]

Broken into #1060, #1061, #1062, #1063, #1064, #1065, and #1066.