Add support for running with IOSvc

[jmcarcell]

BEGINRELEASENOTES

• Add support for running the wrapper with IOSvc with a few tests:
  • Run the CLIC reconstruction with IOSvc (proves processors run under IOSvc, producing the same output).
  • Run global_converter_maps with IOSvc, where a functional algorithm produces output that is used by a Marlin processor and the output of the two is used either by a Gaudi::Algorithm or a functional algorithm (proves processors can take input from functional algorithms and functional algorithms can take input from processors).
  • Run test_link_conversion_edm4hep with IOSvc, where links produced by functional algorithms are checked by Marlin processors.

ENDRELEASENOTES

• Needs Add the possibility to set edm4hep::utils::ParticleIDMeta via the IMetadataSvc k4FWCore#273
• Needs Add support for reading collections stored as DataWrapper<podio::CollectionBase> in functional algorithms k4FWCore#280

Fix #202

[tmadlener]

Maybe this also needs a similar treatment as convertObjectParameters in k4EDM4hep2LcioConv? That would make it possible to simplify the code below a bit, but I am not sure if it's worth the effort.

[andresailer]

Could we have a one line description for each service under what circumstances which service is used?

[jmcarcell]

Like the ones I added?

[jmcarcell]

I just realised I made a mistake and key4hep/k4EDM4hep2LcioConv#102 is not needed, so it can be reverted. key4hep/k4EDM4hep2LcioConv#102 was needed before because I was incorrectly writing the event parameters to the metadata frame. I discovered a limitation (added in a warning) of the current implementation: when running Marlin processors with IOSvc without any input file, it's not possible to convert event parameters to EDM4hep as the frame that will be written doesn't exist yet; it will be created later by the Writer: https://github.com/key4hep/k4FWCore/blob/main/k4FWCore/components/Writer.cpp#L200.

[tmadlener]

when running Marlin processors with IOSvc without any input file, it's not possible to convert event parameters to EDM4hep as the frame that will be written doesn't exist yet; it will be created later by the Writer:

This is mainly (only?) relevant for the LCIO to EDM4hep direction, right? As far as I can tell, the tests are unchanged apart from the additional infrastructure to switch between IOSvc and PodioDataSvc. In that case I would say this is OK for now and we check again in the future if someone runs into problems with this.

[jmcarcell]

I have edited the first post with information about all the tests. I think after key4hep/k4FWCore#280 functional algorithms and IOSvc should play well with the Marlin wrapper unless I'm missing any situation that is not in the tests...