An empty block with height is often considered as containing a visible position. Example:

<div contenteditable></div>

The logic of identifying such positions is replicated at several places in the code base, each with its own ad-hoc implementation:

VisibleUnits.cpp
- EndsOfNodeAreVisuallyDistinctPositions()
- IsVisuallyEquivalentCandidate()

VisibleUnitsLine.cpp
- StartPositionForLine()
- EndPositionForLine()

We should unify them to make our code base healthier.