ATR-19246 TriggerMT Decision Encoding in ESD Update

This persists in the trigger output a HLTSummary, and then in the ESD a xAOD::TrigDecision object the bit-set which corresponds to fired triggers in the new MT prototype. cc @tbold @bwynne

It hacks a short-term hard-coded fudge into the trigger configuration service in order to obtain ChainCounters. cc @mark @fwinkl @stelzer

• Remove my recently added extension to the TrigDecision EDM. Disk-space wise, this was not going to work, sadly.
• Update downstream clients of DecisionSummaryMakerAlg within the trigger (signatureMoni, streaming, bit-encoding in HLTResult)
• Update TrigDecisionMakerMT to read all HLT summaries from one collection, DecisionSummaryMakerAlg's HLTSummary
• Update transient HLT seeding information written by L1Decoder, call this L1DecoderSummary everywhere.
• Copy TrigCost end-of-event trigger fire to DecisionSummaryMakerAlg
• Remove the output from the TriggerSummaryAlg - this now just prints to terminal
• Hack the TrigConfSvc to allow hard-coded chains to be injected in the short term, for testing.

(note known problem with egamma test, will try and fix that now too - but in its own MR)