[ehellbar]

This allows to store the estimated NotAfter timestamp of a period and pass in
CCDB as a CCDBTimeMachineTS object, so that the estimate is done once per
pass and not in every job.

• The script first queries CCDB for an existing object for the Period and
  Pass, and only estimates from the catalogue when none is found.
• --publish-to-ccdb uploads the estimate, --force-overwrite replaces an
  existing object and --skip-ccdb-query skips the lookup.
• The CCDB cross-check now uses the run start timestamp.
• Only the value goes to stdout; all log lines go to stderr.

Nothing uses the script in production yet.

https://its.cern.ch/jira/browse/O2-6543

[github-actions]

REQUEST FOR PRODUCTION RELEASES:
To request your PR to be included in production software, please add the corresponding labels called "async-" to your PR. Add the labels directly (if you have the permissions) or add a comment of the form (note that labels are separated by a ",")

+async-label <label1>, <label2>, !<label3> ...

This will add <label1> and <label2> and removes <label3>.

The following labels are available
async-2023-pbpb-apass4
async-2023-pp-apass4
async-2024-pp-apass1
async-2022-pp-apass7
async-2024-pp-cpass0
async-2024-PbPb-apass1
async-2024-ppRef-apass1
async-2024-PbPb-apass2
async-2023-PbPb-apass5

[sawenzel]

Thanks, this is a good step towards https://its.cern.ch/jira/browse/O2-6543. Apart from the small inline points, I would merge it as is, since nothing uses the script in production yet.

One question for the next iteration: the object is keyed by Period and Pass, but the value is estimated from a single run. Do we know that all runs of a pass are reconstructed close enough in time for one value to fit them all? If not, we could key by run, or have each job compare the shared value with its own estimate and warn when a reference object like TPC/Calib/CorrectionMapV2 resolves differently.

Next steps: publish and consume the object from o2dpg_sim_workflow_anchored.py, following MCProdInfo, with ALIEN_JDL_CCDB_CONDITION_NOT_AFTER kept as the override. The certificate and upload helpers could then move to a common module, as your FIXME already suggests.

[ehellbar]

@sawenzel thanks for the review. I pushed two more commits which should address your suggestions.