Add an input-management layer

Currently, keyboard and mouse input is all handled in an ad-hoc manner in window.rs. Adding an extra layer of abstraction could help simplify the process of adding a new control. We could solve the following problems in one central place (although we don't need to solve them all now to resolve this issue):

• Allowing rebinding of controls
• Adding contextual controls
• Having input presses that trigger an event on the next tick
• Having an axis control for movement
• Detecting conflicts (and allowing the same key to do multiple things if unambiguous enough)
• etc. (This list is not exhaustive.)

Ideally, after this issue is done, if somebody wants to add a new key that does something, they'll just need to add the logic with minimal boilerplate code.

Start by reading window.rs and inventorying how keyboard and mouse input are currently handled. Narrow the broad list of possible goals to a defined first increment, then verify that adding a new control requires minimal boilerplate and that the chosen behavior is covered by the project’s existing checks.