Is it worth it to normalize?
Let me give a different answer here:
First of all, let's state that 'normalizing' is a destructive signal processing task that can be done during the mastering of a 2 track mixdown.
In general: Try to avoid normalization as much as possible. Or, in better words: try to avoid the need for it. Try to record in such a way that you don't let your converters clip, but you keep the headroom as little as possible. Make sure your pre-aplifiers are gained appropriately. If you mix down multiple channels that you recorded loud enough, you should practically never end up with a signal that is too soft on the master.
This starts with a decent soundcheck. Make sure your recording signal approaches as close to 0dB FS as possible, but without the occurrance of clipping.
In my opinion, after the mixdown, headroom of > 1dB is unacceptable. No matter what type of music you make. So before you introduce the master, you should already be aware of the total sound level. (And you should rather get that done using the mixer, than by using some gain plugin at the master.)
If for instance you record classical music in multiple tracks, make sure that all the microphones are giving a decent signal when the musicians are playing. If you level your main microphones correctly, you should always have a bit too much sound level rather than a bit too little. In that case, at the moment you mix down, you can either lower your master fader just a bit (I normally get it between -1 and -2 dB attenuation on a 24 tracks recording of a symphony orchestra) or use a compressor to squeeze the dynamic range a bit.
Because normalizing is an excuse for not using the entire availabe dynamic range. If my recording is 16 bits (equal to 96 dB dynamic range) and at the mastering stage, I have to gain the master track 12 dB, you immediately diminish the dynamic range to 84dB, which is an audible loss of quality, especially in music with much dynamic range. If you find yourself in this situation, instead, alter your mix and re-bounce the audio, so that you get away with a master that has only < 1 dB headroom.
So, if the audio needs to be normalized it is worth it and would be considered constructive.
But, it is often combined with other destructive processes, for example:
Depends on what you want to do. For ripping old audio CDs for your music collection, do what you want... :->
For a recording scenario, this is surely not feasible. If you normalize every song, you'd have different average loudness per track. You surely do not want that. If you normalize single instrument tracks, you'll win nothing, because you're going to do a lot of mixing, compressing and mastering anyway.