[Feature Request] Define canonical inward direction for surfaces

[bbarad]

Pseudocode plan:

1. Measure the center of mass of each section of surface when it is first generated (or of the overall? gotta test)
2. For each isolated section, test which flip minimizes the area-weighted average of the angle between the normal and the line to the center of mass.
3. Choose the flip which minimizes.

[bbarad]

My current plan is to work on this for limited patches, by orienting relative to a desired surface - aka definining "in" as towards the partner in an organelle contact patch. We are going to prioritize the generalized solution in the next few months.