Add a HLT1 filter to the Hlt2SingleHighPtMuon lines
This MR does the sensible thing of applying a HLT1 filter on the relevant HLT2 signal SingleHighPtMuon lines. This has a knock-on effect to Sprucing, which takes directly the output of the HLT2 lines.
Inspired by !4290. Recent studies from @lugrazet have shown that there is some junk that our HLT1 line doesn't fire on, but gets through via TIS on other lines and then our HLT2 lines let it through. We think this could be primarily ghosts at high eta - we know HLT2's ghost rate is worse than HLT1, particularly at high eta.
Rate and bandwidth impact
Following the QEE MR policy: tested with https://gitlab.cern.ch/-/snippets/3438 on 50k HLT1-filtered input events. I made no changes to any Turbo lines, so switched off that stream in my local running. The reference is the qee_upgrade branch on 0ae2c8aa and the stack is master compiled on the morning of 12/03/24 on the default platform.
| Line/Stream | Rate Before (Hz) | Rate After (Hz) | Rate Change (Hz) | BW before (MB/s) | BW after (MB/s) | BW change (MB/s) |
|---|---|---|---|---|---|---|
| Hlt2SHPMuon | 620 | 340 | -280 | - | - | - |
| Hlt2SHPMuonIso | 340 | 150 | -190 | - | - | - |
| Hlt2SHPMuonPrescale | 78 | 50 | -12 | - | - | - |
| Hlt2QEE Full Stream | 5100 | 4800 | -300 | 700 | 620 | -80 |
| SpruceSHPMuon | 310 | 130 | -180 | - | - | - |
| SpruceQEE Stream | 1500 | 1300 | -200 | 120 | 85 | -35 |
In conclusion: Reduces Hlt2 QEE full stream rate (bandwidth) by around 300 Hz (80 MB/s) and QEE sprucing rate (bandwidth) by around 200 Hz (35 MB/s).