Define ContentGraph memory cache flushing

[mhsdesign]

good point, this is a dealbreaker IMO – probably it makes sense to call this something else (e.g. resetRuntimeCaches) and then reset the inMemoryCache (we should not disable it).

The main goal of this change was not to expose all those nasty internals such that we can refactor more easily.
In fact inMemoryCache is now only public because it is still used by the nasty code of the frontend controller (which I plan to clean up)

But we should make sure that:

$subGraph->find*(...);
$contentRepository->handle(...);
$subGraph->find*(...);

behaves as expected, i.e. the second subgraph interaction contains the changes.

Lets see if the tests fail – and if they don't we should add scenarios that reflect the above!

Originally posted by @bwaidelich in #5028 (comment)

[mhsdesign]

points from todays meeting

        * long running process where other process changes content repository? How to flush
            * manually call markstale??
        * flush in same process when mark stale
        * disable vs flush confusion?

[bwaidelich]

I think this issue is obsolete with #5243 and can be closed

[mhsdesign]

Yes was about to look this up when i read your issue but you found this already ^^

I believe the only thing that this covers also is the runtimeCache handling in the workspaceFinder which is not incorporated in your issue?

[bwaidelich]

I believe the only thing that this covers also is the runtimeCache handling in the workspaceFinder which is not incorporated in your issue?

That's right, but the workspace projection will be incorporated into the content graph projection anyways

[mhsdesign]

ahhh yes i forgot about that todo ;) so +1