When InnoDB monitor output to the error log is enabled due to long semaphore waits, the original innodb_status_output setting is preserved, and restored when the auto-monitoring phase ends.

With "67% lock heaps" on the other hand, innodb_status_output is just set to false, regardless of the original setting.

So in the later case, if a user had explicitly enabled innodb_status_output on the SQL variable level, such setting would honored in the long semaphore wait case, but not the 67% lock heaps case.