Add topo and cut-based triggers to inclusive selection of BuToHpLL and B0ToHpHmLL, and remove obsolete spruce lines for these decays

Part of a consolidation of RD trigger lines

What's done:

• Add topo and cut-based triggers to the trigger selection of the BuToHpLL and BuToHpHmLL sprucing lines
• Remove obsolete Spruce lines
• Remove electron lines in HLT2 that are already in Spruce
• Moved code from b_to_xll_spruce_hhll.py to b_to_xll_spruce.py
• Tightened PIDe cut from 1 to 2, as the IDD and cut-based triggers already have this cut in place and there is no point in having a looser cut (but topo trigger does not have this cut)
  • During the hackathon this cut was loosened to 0 (to allow for pass-fail in analysis), but the rate increased too much
• Add calo_digits=RD_PERSIST_CALO_DIGITS and calo_clusters=RD_PERSIST_CALO_CLUSTERS to all electron lines in spruce
• Added long-upstream dielectrons to the BuToHpEE and BuToHpHmEE lines in b_to_xll_spruce.py
• Added ECAL acceptance cuts to all dielectron lines in b_to_xll_spruce.py
• Changed all rho0 builders to kstar0 builders in b_to_xll_spruce.py

HLT2 rate tests

The HLT2 rate is checked by running over 100000 events from 2024 data. Before these changes, the HLT2 rate was 17.0 kHz in HLT2. After the changes, the total rate is 15.8 kHz, which is expected since many lines in HLT2 were deleted.

Spruce BW tests

For sprucing, the bandwidth was estimated by running over 100000 events from EoY sprucing. Before the changes, the output file is 64 MB and the output contains: LAZY_AND: moore #=59741 Sum=2532 Eff=|( 4.238295 +- 0.0824243)%|

Assuming an input rate of 75 kHz, the bandwidth is therefore (75000*(64e6/2532)*2532/59741)/1e9 = 0.080 GB/s.

After the changes, the output is: LAZY_AND: moore #=59741 Sum=2891 Eff=|( 4.839223 +- 0.0877972)%| The output file is 91MB, so the bandwidth increases to (75000*(91e6/2891)*2891/59741)/1e9 = 0.114 GB/s. So the rate has increased by 14%, while the bandwidth has a 42% increase.

To mitigate the increase in bandwidth, the cuts in dielectron lines were tightened, as these lines retain many high-multiplicity events. Specifically, the Kst and dielectron vertex chi2 and DOCA chi2 cuts have been tightened to 10.0. In SpruceRD_BuToHpEE the hadron IP chi2 cut has been tightened from 6.0 to 8.0, while in SpruceRD_B0ToHpHmEE it's been tightened from 6.0 to 9.0. The last change also removed a warning about too many combinations.

With the same calculation as above, the bandwidth is now (75000*(67e6/2427)*2427/59741)/1e9 = 0.084 GB/s. The output is

LAZY_AND: moore #=59741 Sum=2427 Eff=|( 4.062537 +- 0.0807712)%|

In summary, the rate has dropped by 4%, while the bandwidth has increased by 5%, compared to before this MR.