Refactor: background canvas layer creation in simulator

[psychemedia]

Creating a new canvas layer in the simulator is a faff; it would be better to use a function to create a new canvas; that way, we can more easily add new layers for different sensor types, where the sensor responds visually, for example, to things on that particular layer.

This means we can model detection of a fixed light sensor or IR beacon, for example, using a visual field on a hidden layer.