Remove sync workaround in GPUBuffer.map_async()

[almarklein]

The workaround:

wgpu-py/wgpu/backends/wgpu_native/_api.py

Lines 2556 to 2562 in 22a1e3d

	# Sync up when reading, otherwise the memory may be all zeros.
	# See https://github.com/gfx-rs/wgpu-native/issues/305
	if sync_on_read and map_mode & lib.WGPUMapMode_Read:
	if self._mapped_status[2] == 0 and self._usage & flags.BufferUsage.MAP_READ:
	encoder = self._device.create_command_encoder()
	command_buffer = encoder.finish()
	self._device.queue.submit([command_buffer])

Links:

• Same issue in wgpu-native: Reading mapped buffer data yields zeros when the data is written using the mapping API gfx-rs/wgpu-native#305
• The upstream fix: fix(core): Flush pending writes before mapAsync, if needed gfx-rs/wgpu#9307