Task Description

Repo process should generate one script per call, just like worker execute. This also makes the process more debuggable.

Tasks

[x] Migrate TODOs from old implementation to new one.
[o] Implement net_splitcells_os_state_interface_repo_process_repo_peer_disabled -> This is not needed, as it is not used and because the --ignore-peer-repos is properly propagated.
[x] Implement current processing.
[x] Implement children processing.
[x] Implement peer processing.
[x] Implement recursion.
[x] Implement missing repo processing.
[x] Implement unknown repo processing.
[x] Implement optional parallelism.
[x] Implement --ignore-peer-repos.
[x] Pass repos process calls to new implementation via environment variable.
[x] Make child repo naming consistent in script. See sub vs child repo.
[x] Deprecate old repo process and add task for the future to replace delete the old process. Thereby, the migration would be finished. -> It's done the other way around: v2 is used by default and only a special flag enables v1. This way v2 is tested as much as possible via normal usage. The v1 integration in v2 has an end date: if the 2026.03.12 is reached, any execution of v2 will fail and require the removal of v1 and its integration in v2 from source code.
[x] Format multi line string example, to not mix up code indentation.
[x] Log each processed repo in generated script.
[x] Test high nesting level like on developer computer.
[x] Make repos.push.at.all work.