Tissue masker on multiple images

[measty]

Someone has commented in an old PR, not sure if they intended to make an issue instead, but it seems like a valid issue so making one here.

Seems like masks should be returned instead of [mask]. Otherwise, the transform doesn't truly process a list of images.

Originally posted by @samokhinv in #61 (comment)

To elaborate:

tiatoolbox/tiatoolbox/tools/tissuemask.py

Lines 143 to 167 in e899cec

	def transform(self: TissueMasker, images: np.ndarray) -> np.ndarray:
	"""Create masks using the threshold found during :func:`fit`.
	Args:
	images (:class:`numpy.ndarray`):
	List of images with a length 4 shape (N, height, width,
	channels).
	Returns:
	:class:`numpy.ndarray`:
	List of images with a length 4 shape (N, height, width,
	channels).
	"""
	super().transform(images)
	masks = []
	for image in images:
	grey = image[..., 0]
	if len(image.shape) == 3 and image.shape[-1] == 3: # noqa: PLR2004
	grey = cv2.cvtColor(image, cv2.COLOR_RGB2GRAY)
	mask = (grey < self.threshold).astype(bool)
	masks.append(mask)
	return [mask]

The problem with this function that the commentor highlighted is that if you give a list of images you are only going to get out the last one processed.