Add min_size_in_mm to RemoveSmallObjects transform

[AHarouni]

Describe the solution you'd like
RemoveSmallObjects is a great transform that is very much needed. However its arg is
min_size (int) – objects smaller than this size (in pixel) are removed.
Tumors and organs in medical are measured in mm^3 or CC^3. As radiologist same remove tumors that are smaller than 3 mm

Describe alternatives you've considered
I had to use a transform to resample to 1x1x1 so the pixel count is 1mm^3 remove the tumors then resample back to org size. Instead this should be a simple multiplication as we have the dims actual measurements in the metadata

[wyli]

the min_size can take into account the metatensor.pixdim if it's a MetaTensor before the converting

MONAI/monai/transforms/utils.py

Lines 1089 to 1090 in cc3faa3

	img_np: np.ndarray
	img_np, *_ = convert_data_type(img, np.ndarray)

MONAI/monai/transforms/utils.py

Line 1101 in cc3faa3

out_np = morphology.remove_small_objects(img_np, min_size, connectivity)