This was cool, but it's really hard to optimize the keymap with this feature because we never know if a keystroke will match against a binding set with a function, which will force us to always consider this binding set against every key event.
Certain events call `abortKeyBinding` to opt out of handling certain keybindings. Snippets does this with tab for example. If it's not a situation where it's appropriate to go to the next tab stop, we let the next binding be triggered, which could insert a tab, for example. But when we trigger events from the event palette, there *is* no next binding. Having a no-op function helps in this situation.
For example, if you have a very unspecific binding "ctrl-x ctrl-c", you can also have very specific "ctrl-x …" bindings, and bindings from both sets can be matched. A partial match in a more specific does not rule out later matches in a less specific set.