Kubernetes memory requests and limits fixed-point vs power-of-two

Question

There are two examples on kubernetes.io about how to specify memory request and limits and they are very confusing.

(I removed cpu and the unnecessary quote