* Implement, test container-over-container broadcasting

* Tweak bcast_container_types arg doc

Co-authored-by: Alex Fikl <alexfikl@gmail.com>

* Improvements to with_container_arithmetic

- Support matmul with a separate operator class
- Join container and number broadcasting code paths
- Make args keyword-only

Co-authored-by: Alex Fikl <alexfikl@gmail.com>