Should map reuse permit maps with larger max_entries?

[Alan-Jowett]

Currently when reusing an existing map while loading a BPF program the maximum entries in the map is required to match.

Is there a compelling reason to keep this requirement?

Scenario that is blocked:
BPF program that is offline verified with a map with some default initial size.
At runtime, the application determines the actual required map size and creates a map of that size.
At load time, the BPF program is used with the new map size.