Don't try to block commands while uploading a level

[Toastbrot236]

This gets rid of the mechanism which uses CommandService to block commands while publishing a level.

We can't catch all cases of /filter requests which are not chat messages (e.g. LBP3 sends the level strings to filter before its /startPublish request, and LBP2 can also send level strings during create mode).
More importantly, if the game successfully does a /startPublish request, but its /publish request fails in any way or the user tries to update a level instead of publishing a new one, their commands would continue to silently get blocked, which has already caused confusion (see #937).

The latter issue could be fixed and worked around by making some changes to the /publish endpoint method and the CommandService, but since we also can't tell whether a /filter request does actually contain a chat message or something else half the time, making the mechanism not very effective to begin with, I think it makes more sense to just get rid of it entirely.

Closes #937.

[jvyden]

Yeah this system never really worked.

This was done for security reasons (theoretically commands could be injected as you're publishing a copied level) but it's fine as we explicitly do not do trusted operations in filtering.