1) "Normal" is what's normal to you. People invest way too much energy in apparent volume (and far too much volume). If it's not loud enough for you, hit it harder. If it's too loud, pull back. If your monitoring chain & controller isn't properly calibrated against a set reference level, it's going to be an uphill battle from the start.
2) A "Maximizer" is basically exactly the same as a compressor - not the opposite. It makes loud things quieter. The opposite is an expander - It makes quiet things quieter. The "extreme" version of a compressor being a brick-wall limiter / maximizer. The extreme version of an expander being a noise gate. Volume automation would be your friend if you're trying to make quieter parts louder - But I'd argue that's for the mixing stage if they're your own mixes. Saving that for mastering is like icing a cake before you bake it.
3) I'm not exactly sure what you mean there.
4) It changes for every track, but the general process would be correction first, EQ, dynamics. But that's just a rule of thumb.... But no means a hard rule.
I do most corrective processes in digital, some EQ in digital, anything like a dynamic EQ or de-essing would be digital, *some* dynamic control (whether that's volume automation or light stereo or mid-side compression) in digital, then the final shaping EQ and "color" comp is analog on the way back in at the "sweet spot" volume (the volume that actually best serves the music). Then usually some sort of digital limiting for "commercially acceptable" / CD levels. But again - Some tunes go through incredibly light amounts of processing, others incredibly heavy amounts. The tune tells you what it needs. If it's not telling you, that's another issue anyway. But as far as final playback level is concerned, that's totally a personal taste thing.
All that said - If you're making beats that are to be used in the creation of songs, NO limiting or excessive compression and a decent amount of clear headroom. If you want to make a "squashed preview" version so people might hear what it might sound like at "commercially appropriate" levels, that's fine. But if you're selling / leasing said beats, you're going to piss a lot of people off if they've been dynamically compromised.