Split-up "Rework CR CatchUp mechanism"

[mhsdesign]

As the PR #4988 will likely NOT land in the final 9.0 we have to extract the important bits and peaces were we told ourselves yeah that will be resolved with this PR.

Ill start:

• we continue to have a CommandResult as return value of ContentRepository::handle() because we could not decide yet if we need one for example to provide the current sequence number or something. I would be fine with removing and prepared a full removal TASK: Remove references to CommandResult in return types #5061 which was extracted from Bastis rework
  • TASK: Remove references to CommandResult in return types #5061
• The catchuptriggering with synchronous mode is also hacky hardcoded and we can get rid of the interfaces and the needless abstraction. Cherry pick fc6ec5c
  • TASK: Cleanup projection catch-up trigger extensibility #5288
• Remove doctrine migrate and cr setup commands from behat scripts and avoid "A transaction is active already, can't commit events!" BUGFIX: Use dedicated connection for EventStore #5005
• update neos docs to not promote block see https://docs.neos.io/guide/contributing-to-neos/event-sourced-content-repository/blocking-until-a-command-is-executed and related
• Proper(er) mechanism to prevent multiple catchup runs TASK: Prevent multiple catchup runs #4751?

  • I guess that is possible, although without async I am not sure there even is a case anymore where this can happen? Probably can with multiple users in parallel so mmmm yeah

• move resetProjectionStates and resetProjectionState out of CR
  • !!! TASK: Declare ProjectionService as api to reset projections instead of ContentRepository #5292
• catchup logic with hooks into event persister / catchup logic without hooks? in site(replay)service?
• optimise catchupAllProjections to not loop over each projection but load all the required events instead and then apply on projection
• apply on each projection directly without can-handle (that would also allow catchup hooks to be run always?)
• only catchup projection if it is already at the highest sequence number? And introduce cli commands to catchup and cr:status to notify if a catchup is needed
• simplify checkpoint storage implementation by just returning the highest applied sequence number
• Ensure exceptions during CatchUpHook are properly handled #4908

[mhsdesign]

all these points are solved - mainly implemented via #5321