[Feature] Root volume per mm of depth

[tparkerd]

Is your feature request related to a problem? Please describe.
Implement trait for "root volume per mm of depth".

Describe the solution you'd like
For each millimetre of an XRT volume, count the number of white pixels in each binary image that represents a slice of volume, and multiply it by the slice thickness for each dimension: x, y, and z.

To determine the number of slices that represent 1 millimetre of depth, 1000 µm ÷ thickness_z = N_slices.

For example,

1000 µm ÷ 109 µm ≈ 9.174 slices

If the volume is downsampled by half, then the thickness changes.

1000 µm ÷ 218 µm ≈ 4.587 slices 

This may result in unexpected numbers depending on which side of the slice you want to include.

• Should the calculation be greedy and assume that if the slice falls without the bounds, include it?
• What happens if pixels would be counted for both a 1 mm subset and the following one?
• Would using a greedy method cause volume to spike at the boundary between two sections?
• Should it be sequential calculation where you sum the volume for all the slices in a volume until it no longer meet the condition? (E.g., Sum values for 0 ≤ x < 1)

Additional thoughts
During coding, sum the total number of pixels in the volume and check to see if the sum of all the pixels is equal to the count of pixels used in the calculation of the trait.

For example:

For all pixels, P, in each 1 mm section, (where n = count of 1 mm sections in volume)
P₁ + P₂ + P₃ + ... + P_n-1 + P_n = P_volume

For each all voxels, V, in each slice, (where n = count of slices in volume)
V₁ + V₂ + V₃ + ... + V_n-1 + V_n = V_volume

Check if P_volume = V_volume

If they are not equal, there is data loss or noise.