…delete

the logic is carried out inside `RebaseableCommands::commandMatchesAtLeastOneNode`

The reasoning behind this is that asking the commands which node aggregate id is affected _does_ work but the capabilities stop if we want to ask which dimensions are affected.

For that we must use the event instead - at least in the future if we want to reintroduce dimension based publishing again. The matching list from commands to node id will also be simplified once we use the `EmbedsNodeAggregateId` interface instead and operate on events.

…stead of `NodeIdsToPublishOrDiscard`

... as a 3rd option the followup commit will just flush the full asset usage workspace!

…arded in the events (for now)

Introduced with neos@f2c66e0 - its not important for the graph projection and only used by the asset usage catchup hook which can be adjusted.

If we want to re-introduce `$publishedNodes` to the event should be also careful that we calculate them from the _actual_ events so we can know for sure there is no gibberish in the event.
Previously for example discarding a non-existing node or a node in a not valid DSP would just save that directly into the event, which is not correct as that did not happen like that. It was merely a bad intention.

Instead of fine granular updating only the usage references on the nodes that were discarded, well throw away everything and then rely on the reapplied events what is kept.

This is more consistent to how other parts react to publishing like the graph just forks a new content stream and doesnt care about the old data, the change projection behaves similar to the graph.
And for the content cache flusher we had the discussion to use `discardedNodes` but opted against it and do a full flush there as well: neos#5175 (comment)

This change allows us to get rid of the `discardedNodes` property (with dimensions) in the `WorkspaceWasPartiallyDiscarded` event.

reasoning:

- the `publishedNodeIds` were removed on the event, as we dont need it, and it would rise a discussion if we need to keep dimension space points there too or not: see neos#5386 how i archived it.

- instead of now trying to make `WorkspaceWasPartiallyPublished` contain all the information we dont use or need, the idea is to make it dumb. Now its basically just a `WorkspaceWasPublished` and then the other remaining events are applied as if they were carried out by a user.
If we need to add more metadata to the partial publish at some point, that would need a new event id say as we woouldnt be able to mess with the `WorkspaceWasPartiallyPublished` once released.
So by just combining them into one we have the option open to create a `WorkspaceWasPartiallyPublished` again but differently which wil be breaking too but differently. Its just a new event then, no need to migrate data.

…d WorkspaceWasPartiallyDiscarded

…ross-all-dimensions

…e dimension for now

Reverts d2d9faa partially

Initially this change was done to fix neos#5303
But that was a hoax see neos#5303 (comment)

we always emit the `WorkspaceWasPublished` event and this just simplifies the code-path again.

For the discard all case, the code is stil a little easier to read _with_ the condition.

…t runtime

We cannot erase history just because we want to.
Especially because this does not fix any bugs.

…d spread around

This overhauls the `NodeAggregateWasDisabled` upcasting from

neos#4659

This was done do ease debugging and still preserve the intention.

…nds for match