nonmoving: Optimise log2_ceil

This is both simpler and produces slightly better (branchless) machine code on x86.

Edited by Ben Gamari

Merge request reports

Loading