Allow reading back from canvases after present

Changes compositingAlphaMode to just alphaMode, because it now affects
the canvas as an image source (drawImage/toDataURL/etc.) as well as
compositing.

Fixes gpuweb#2743
Fixes gpuweb#1847
Fixes a leftover bit from gpuweb#2373 (placeholder canvases)

explainer/index.bs

@@ -1040,15 +1040,17 @@ even though context state does not persist across loss/restoration.)
 In order to access a canvas, an app gets a `GPUTexture` from the `GPUCanvasContext`
 and then writes to it, as it would with a normal `GPUTexture`.
 
-### Swap Chains ### {#canvas-output-swap-chains}
+### Canvas Configuration ### {#canvas-output-swap-chains}
 
 Canvas `GPUTexture`s are vended in a very structured way:
 
-- `canvas.getContext('gpupresent')` provides a `GPUCanvasContext`.
-- `GPUCanvasContext.configureSwapChain({ device, format, usage })` provides a `GPUSwapChain`,
-    invalidating any previous swapchains, attaching the canvas to the provided device, and
-    setting the `GPUTextureFormat` and `GPUTextureUsage` for vended textures.
-- `GPUSwapChain.getCurrentTexture()` provides a `GPUTexture`.
+- `canvas.getContext('webgpu')` provides a `GPUCanvasContext`.
+- `GPUCanvasContext.configure({ device, format, usage })` modifies the current configuration
+    invalidating any previous texture object, attaching the canvas to the provided device,
+    and setting options for vended textures and canvas behavior.
+- Resizing the canvas also invalidates previous texture objects.
+- `GPUCanvasContext.getCurrentTexture()` provides a `GPUTexture`.
+- `GPUCanvasContext.unconfigure()` returns the context to its initial, unconfigured state.
 
 This structure provides maximal compatibility with optimized paths in native graphics APIs.
 In these, typically, a platform-specific "surface" object can produce an API object called a
@@ -1057,14 +1059,15 @@ into.
 
 ### Current Texture ### {#canvas-output-current-texture}
 
-A `GPUSwapChain` provides a "current texture" via `getCurrentTexture()`.
+A `GPUCanvasContext` provides a "current texture" via `getCurrentTexture()`.
 For <{canvas}> elements, this returns a texture for the *current frame*:
 
-- On `getCurrentTexture()`, `[[currentTexture]]` is created if it doesn't exist, then returned.
-- During the "[=Update the rendering=]" step, the browser compositor takes ownership of the
-    `[[currentTexture]]` for display, and that internal slot is cleared for the next frame.
+- On `getCurrentTexture()`, a new `[[imageData]]` is created if one doesn't exist for the
+    current frame, wrapped in a `GPUTexture`, and returned.
+- During the "[=Update the rendering=]" step, the `[[imageData]]` becomes readonly. Then, it is
+    shared by the browser compositor and the page's canvas (readable using drawImage/toDataURL/etc.)
 
-### `getSwapChainPreferredFormat()` ### {#canvas-output-preferred-format}
+### `getPreferredCanvasFormat()` ### {#canvas-output-preferred-format}
 
 Due to framebuffer hardware differences, different devices have different preferred byte layouts
 for display surfaces.
@@ -1089,8 +1092,8 @@ As today with WebGL, user agents can make their own decisions about how to expos
 capabilities, e.g. choosing the capabilities of the initial, primary, or most-capable display.
 
 In the future, an event might be provided that allows applications to detect when a canvas moves
-to a display with different properties so they can call `getSwapChainPreferredFormat()` and
-`configureSwapChain()` again.
+to a display with different properties so they can call `getPreferredCanvasFormat()` and
+`configure()` again.
 
 #### Multiple Adapters #### {#canvas-output-multiple-adapters}