I´ve tried making a Mario-clone and I can atest that doing something more self-contained, like Celeste or Super Meat Boy, is a lot more simple. For starters, you can save a lot of time repeating assets and reusing mechanics: as challenges are so small and quick you can use a new type of platform in several challenges before it becomes too tiring, but in a Mario-like game, where levels are much longer and the player has to replay parts of it everytime he dies, that can easily become repetitive. Also, every level must present a new mechanic, so in the end you have to design a lot more stuff and you can´t really use it to create as much content as in "One-screen-challenge-platformer".
The same goes for assets: you can spend quite some time in a level if you try to explore and take your time, so reusing backgrounds, music, etc...will soon become much more apparent than in something like Meat Boy, where you are constantly changing screens and are always advancing towards a new level with a new setting.
A very important aspect is that it´s just much more complex to edit a big level than a small one. Even in big levels, every platform is designed with intent behind almost every pixel, so when you find that a challenge in the middle of the level isn´t working like you expected and you need to reallocate it or make it use more/less space, it may also force you to redesign a lot of other parts of the level. In one-screen-challenge-platformers the levels are tiny in comparison, so any modification is much easier to implement. If you try to put shortcuts and secret areas in the level the complexity scalates like crazy, though this can be kept more manageable if you make those areas isolated like the bonus areas in Mario which are entered through pipes.
Big levels have other associated nuances as well, like the engine having to manage a big level in real time instead of just a small room where everything can be controlled much more easily. You have to control when enemies and platforms start to move so you don´t screw the level´s pace, you have to manage the camera to frame correctly each challenge when the player approaches them, you have to take care that something off-camera doesn´t screw the whole level, like an enemy with a required key falling on a pit...
Finally, Mario is kind of looked down by enthusiasts. Just look at this thread and people downplaying the New Super Mario games (BTW, NSMBU is one of the best 2D Marios, up there with SMW and SMB3). Most likely a Mario clone will be looked down like a begginner´s game, a game for kids, or something plain uninteresting. However if you come up with a cool gimmick like time manipulation for Braid, you can advertise your game on that and be received much better, almost like a work of art. Or focus on difficulty like Meat Boy and take advantage of the perception of Mario being childish to sell your game as "mature/hardcore". You instantly gain more recognition and goodwill from your possible audience, while using less time/resources than if you went for a Mario clone.
Taking all this into account, the real question should be: why would anyone on their right mid make a Mario clone?
Disclaimer: I love Braid and Meat Boy, if anything on my post seems negative towards them it´s just not my intention. I´m just using them as comparison because they are instantly recognizable by everyone. They are both awesome games.
Naw, there were a few puzzles where you had to let the enemies jump off your head with inverse Mario mechanics later on.
Braid is a puzzle platformer, the focus is heavily in finding the solution to the puzzles, to the point where you can just walk to the end of some levels with no opposition. Mario is about clearing a course; the puzzle component is very light and it relies more in skill. Besides that, Braid takes one gimmick (time manipulation) and plays it with in different ways in every world; it´s true that it also introduces some new elements but all of them are designed around the time mechanic. Mario introduces new mechanics in every level and those mechanics hog the spotlight for as long as they´re present, they aren´t a secondary element to anything like Braid´s additional elements. Braid paids homage to Mario with Goombas, the fake Bowser, piranha plants, the theme of saving the princess and such, but the similarities really stop there and are mostly superficial and done in an almost ironic way, like the reversal of the "save the princess" trope at the end.
