dataset/co3d_dataset.py (4 lines):
	- line 22: from plyfile import PlyData  # TODO: get rid of it after pt3d supports loading normals
	- line 176: # TODO: use concatenation
	- line 184: # TODO: make a function for it
	- line 185: # TODO: don't store K; enforce working in NDC space


tools/camera_utils.py (3 lines):
	- line 8: # TODO: all this potentially goes to Pytorch3D
	- line 19: # TODO: fix indexing in pt3d
	- line 20: # TODO: type hints


dataset/dataset_zoo.py (3 lines):
	- line 24: # TODO from dataset.dataset_configs import DATASET_CONFIGS
	- line 100: # TODO:
	- line 162: # TODO: maybe directly in param list


dataset/types.py (2 lines):
	- line 23: size: Tuple[int, int]  # TODO: rename size_hw?
	- line 50: # TODO: not sure if those should be optional. If we have been able to


evaluation/evaluate_new_view_synthesis.py (2 lines):
	- line 102: )  # TODO: do we need to enforce this?
	- line 359: # TODO: extract those to constants


tools/utils.py (1 line):
	- line 45: # TODO: test it


dataset/scene_batch_sampler.py (1 line):
	- line 18: @dataclass(eq=False)  # TODO: do we need this if not init from config?