Reading through the implementation, I feel this line of comment is not consistent with the actually code, so this is for correcting it.

Implementation of bit-exact resize. Internal calls to linear resize updated to use bit-exact version. (#9468)