How compression auto make-up should work? I think I don't get it

Basically, anytime you encode WAV/PCM audio to mp3 or other lossy formats, the peak levels will have a slight increase. It doesn’t matter if you render directly to mp3 from WaveLab, or render WAV files first and encode to mp3/lossy with another app.

Or what I was getting at is if you are releasing the audio as music on iTunes/Spotify/Amazon/YouTube/SoundCloud etc, the master WAV files will eventually be subject to lossy encoding and the peak levels will increase slightly.

Typically, the lower the bitrate of the lossy files, the more the peaks increase.