As mentioned in #677, the members of XRRenderState[Init] were grouped together because they are applied at the same time. This makes sense for the doubles, but should baseLayer be grouped in with them?

Consider:

• baseLayer is the only interface member in these objects (after Single-canvas inline, drop XRPresentationContext #656).
• The current for XRRenderState says, "values that developers can configure which affect how the session’s output is composited," which doesn't seem to describe baseLayer.
• Layers may eventually become much more complex (see Finalize API shape for supporting multiple layers #670), and it may be problematic to group layers in with XRRenderState or to train developers to think about layers in this way.
  • Supporting or polyfilling (as discussed in Finalize API shape for supporting multiple layers #670 (comment)) baseLayer may be more difficult if baseLayer is set through this mechanism if we want to manage layers at the XRSession level.

One possible solution - without trying to address multiple layers now - would be to add void updateBaseLayer(XRLayer baseLayer); to XRSession. The timing for when the new layer gets applied would be the same as updateRenderState(). (Per #513, we probably don't need to expose the current value.)