allow dispatch to PIL image subclasses

[pmeier]

Bug that was spotted by @vfdev-5:

opened_image = PIL.Image.open("test/assets/encode_jpeg/grace_hopper_517x606.jpg")
loaded_image = opened_image.convert("RGB")

assert isinstance(opened_image, PIL.Image.Image)
assert type(opened_image) is not PIL.Image.Image

assert type(loaded_image) is PIL.Image.Image

F.resize(loaded_image, (17, 11))  # ok
F.resize(opened_image, (17, 11))  # raises TypeError, since we currently only allow exact matches for PIL.Image.Image

This PR removes the restriction that we only walk up the MRO for datapoints.Datapoint and thus enabling an "isinstance" check for PIL.Image.Image.

That also means that we can now also register kernels for subclasses of PIL.Image.Image or torch.Tensor. However, since we don't plan on doing that and users are not allowed to do so

vision/torchvision/transforms/v2/functional/_utils.py

Lines 86 to 90 in 498b9c8

	if not (isinstance(datapoint_cls, type) and issubclass(datapoint_cls, datapoints.Datapoint)):
	raise ValueError(
	f"Kernels can only be registered for subclasses of torchvision.datapoints.Datapoint, "
	f"but got {datapoint_cls}."
	)

this has no other side-effect. It actually simplifies the code.

cc @vfdev-5

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/vision/7835

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 7 New Failures

As of commit e3db48a with merge base c1592f9 ():

NEW FAILURES - The following jobs have failed:

• unittests-windows (3.10, windows.4xlarge, cpu) / windows-job (gh)
• unittests-windows (3.11, windows.4xlarge, cpu) / windows-job (gh)
• unittests-windows (3.8, windows.4xlarge, cpu) / windows-job (gh)
• unittests-windows (3.8, windows.g5.4xlarge.nvidia.gpu, cuda, 11.8) / windows-job (gh)
• unittests-windows (3.9, windows.4xlarge, cpu) / windows-job (gh)
• windows (windows.4xlarge, cpu) / windows-job (gh)
• windows (windows.g5.4xlarge.nvidia.gpu, cuda, 11.8) / windows-job (gh)

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[NicolasHug]

LGTM if green, thanks Philip