Disable the async pipeline by default

[DJMcNab]

The async pipeline is not wired up to do robust readback, so it has few advantages (see #366)

This PR also enables the non-async pipeline's interaction with the GPU profiling (which should make it work on wasm)

@simbleau this might also properly fix #494, incidentally.

[simbleau]

Suggested change

	#[cfg(feature = "wgpu-profiler")]
	self.profiler.end_frame().unwrap();
	#[cfg(feature = "wgpu-profiler")]
	if let Some(result) = self
	.profiler
	.process_finished_frame(queue.get_timestamp_period())
	{
	self.profile_result = Some(result);
	}
	#[cfg(feature = "wgpu-profiler")]
	{
	self.profiler.end_frame().unwrap();
	if let Some(result) = self
	.profiler
	.process_finished_frame(queue.get_timestamp_period())
	{
	self.profile_result = Some(result);
	}
	}

[simbleau]

Could we get more detail here? What is meant by "async wiring" - Is this a limitation of !Send futures, single-threaded wasm, etc. - What exactly is the issue on WASM that we aren't doing async pipelines there?

[DJMcNab]

The thread is #gpu > async on wasm. Broadly, the issue is: if we spawn a future, that needs to keep hold of the scene. But that's incompatible with lifetimes of keeping the scene around (to reuse allocations, etc.)

In theory, we should do this much more carefully, but we haven't yet. That might also be useful in contexts outside of browsers too, allowing multi-threaded scene rendering