Reshape a bit the readout map to allow for errors

The new version of the readout map seems to be working, but is delicate to use as any run that has no link conditions attached to it will see 0 banks being activated, leading to lots of messages and possible segmentation faults.

This MR changes the messages and activates all banks and links by default if no bank is found to be active.

added RTA label

/ci-test

added ci-test-triggered label

• [2023-05-23 14:16] Validation started with lhcb-master-mr#7985
• [2023-05-24 12:49] Validation started with lhcb-master-mr#8003
• [2023-05-26 18:30] Validation started with lhcb-master-mr#8047

added hlt2-throughput-increased label

added hlt1-throughput-decreased label

/ci-test

added 1 commit

• 2fa47767 - Change error to warning

Compare with previous version

removed hlt2-throughput-increased label

added 1 commit

• 5375e9fe - Formatting

Compare with previous version

added 1 commit

• 00abc042 - Changing the warning

Compare with previous version

/ci-test

removed hlt1-throughput-decreased label

added 1 commit

• 525672c7 - Simpler syntax to check if we ignore the conditions

Compare with previous version

/ci-test

Started integration test build. Once done, check the results or the comparison to a reference build.

Throughput Test Moore_hlt2_fastest_reco: 482.7 Events/s -- change of 0.19% vs. reference

Throughput Test Moore_hlt2_pp_thor: 112.3 Events/s -- change of -50.47% vs. reference

The fall in throughput is strange, was not there before.

Throughput Test Moore_hlt1_pp_default: 28130.4 Events/s -- change of -0.57% vs. reference

Precisely a 50% drop means only one of the 2 processes did run during the test. This is confirmed by the logs. It ran at the right speed, so I consider it as good. And I merge.

I had a look and there was a SEGV that we need to follow up. I opened Moore#586 . It doesn't look related to this MR so @lohenry you're good

actually this broke a few Allen and MooreOnline tests as seen in the comparision page.

@lohenry can you please investigate with some urgency?

PS we should really be more careful when merging

actually this broke a few Allen and MooreOnline tests as seen in the comparison page.

In the only column we can sensibly deal with (opt detdesc), it FIXED 3 MooreOnline tests and did not touch Allen DD4hep is for the moment not something I can look at, it's just broken. I suppose this is where we fail.

we should really be more careful when merging

Please be my guest and take over maintainership. It's whenever you want. The sooner the better

• [2023-05-30 20:11] Validation started with lhcb-master-mr#8070

added hlt2-throughput-decreased label

added 1 commit

• e0be94d1 - Formatting

Compare with previous version

added hlt1-throughput-decreased label