This copy of the PR has been rebased to remove some of the odd missteps, experiments, and design changes that were holdovers from the original implementation. As a result, the total PR is much more approachable and easy-to-review, but some of the provenance of who wrote what is less than perfect. I attempted to err on the side of giving more credit where possible than less to hopefully not miss anyone's contributions; however, at least one more pass through the code to ensure credit is appropriately assigned should be made before merge.
This copy of the PR has been rebased to remove some of the odd missteps, experiments, and design changes that were holdovers from the original implementation. As a result, the total PR is much more approachable and easy-to-review, but some of the provenance of who wrote what is less than perfect. I attempted to err on the side of giving more credit where possible than less to hopefully not miss anyone's contributions; however, at least one more pass through the code to ensure credit is appropriately assigned should be made before merge.