HLT1 persistency

We want to be able to write out a raw file for all events that pass HLT1. This would ideally also include decision and selection reports, but that be a second step (we don't have all the maker algorithms we need yet).

Requirements

• MDF (+ DST?) writer
  • Works in single-threaded jobs !221 (merged)
  • Works in multithreaded jobs LHCb!2202 (closed)
• Control flow that runs a writer if at least one line selection passes !221 (merged)
  • For now, we can use an InputCopyStream writer, which works for single-threaded jobs The current implementation of InputCopyStream cannot work as it relies on incidents.
• DecReports writing (map of lines to decisions) !226 (closed)
  • Writer !226 (closed)
  • Raw bank serialisation !226 (closed)
• SelReports writing (map of lines to objects that fired the line, in some representation)
  • Writer !293 (merged)
  • Raw bank serialisation !293 (merged)
• VertexReports writing (summary of reconstructed PVs)
  • Writer
  • Raw bank serialisation

Considerations

In no particular order.

1. After the various reports have been serialised, they need to be concatenated with the event raw bank (DAQ/RawEvent) in order for 'the event' to be shipped off to HLT2. Because we schedule the report makers explicitly, we could consider mutating DAQ/RawEvent by appending the report banks.