Loudness normalization sometimes fails to do its job and completely ignores the normalization process, if the sound is not loud enough. Peak normalizing everything during export from daw fixes this, but I don't really see any reason why pre-peak normalization before loudness normalization shouldn't be standard behavior directly in Wwise. Hard to imagine situation where user would enable loudness normalization and want it to ignore specific sounds.
This isn't major issue, since it's easy enough to peak normalize everything during export from daw, but still a bit annoying behavior - especially for delicate ambiences, if you want to have the audio material in approximately sensible loudness for other purposes.