Scrapping a bad level is sometimes an indisputably better alternative to trying to correct what you already have.
Okay, well, I'm not quite sure exactly what to quote out of your message to directly respond to. But I'm going to read between the lines and read your thesis as "Rush jobs can be later made into good maps."
I'm sure you know what a rush job is from experience, but the gist of a rush job is that putting a finish on the map is more important than the new content being worth playing. I think that, by our arbitrary standards of quality exalted in level design 101, a rush job can meet all the same qualifications of a standard map, if not more: my creative process frequently neglects gameplay.
Rush jobs, however, leave an irreversible framework woven into the core of the map. My favorite example here is the hallway. You can make it tall; you can put some floating platforms in it; and you can even put in a balcony. But at the end of the day, you've still got a hallway.
SpiritCrusher hinted at this, but he didn't really elaborate on it. I'll call it "the wisdom of the player." At least as a level designer, I'm reverse engineering your creative process as I'm playing your map. I'm looking at which shapes formed the room in the beginning, and what you've added inside that beginning room. Generally rush jobs have only the most basic layer showing, the raw framework. More developed maps have a second layer just inside the basic framework, which make the basic framework more interesting. On top of that, you have sector detail.
Okay, that didn't make any sense, so I'll have to use an example. I know my own maps best, so I'm using FFZ1. Go to the first room with water in it, by my count the fourth room. (The reason there's a door right between rooms two and three is because I had at first planned on making that room lead to an outside area, but I thought better of it and started on the room you see today. But you didn't need to know that.) The first thing I started with for that room was one, long sector. A hallway in all its glory. The second was the raised sides. The third was the waterfalls. By the definition I've put out in the previous paragraph, the core framework for this room is a hallway. It looks like a hallway in Doom Builder, and changing it into something else is really, really hard. The second layer, which would distract from the blandness of the hall, is made up of the raised sides and the waterfalls. The raised sides give the map shape, and the waterfalls make the walls more interesting. Everything on top of that, (except maybe the trees) is sector scenery.
And, well, capturing map making in such a crude and primitive procedure devalues the creativity of it. If nothing else, I'm trying to describe my process with language where it normally lies just under the surface of my conscious thoughts.
Point is, improving a rush job seems to be throwing sector scenery on top of a basic framework. Sector scenery is only that polish on top of the map that is already there. By the time you've finished your rush job map and decided to go back and improve it, changing the core framework of one room is hard because it's linked to the framing of the other rooms, and adding any sort of "second layer" (okay, that phrase is totally useless) will seem like an addon instead of part of the room's basic structure. Granted, if you're looking hard enough, everything starts to look like a hallway.
Oh dear, I'm not making any sense now, am I?