If you want to get into a discussion about whats better for what, that all comes down to preference in the end. I prefer working in c2 because I can make things so fast. But, I prefer unity because I can do more. I also think given the amount of games released on unity... you get the idea.
I would classify both the binding of issac and Mario as fairly straight forward games, and yes, you could make exact clones. However, here is the thing; You are going to need to write your own collision detection and resolution code to get an exact mario clone. Which brings us to some of the problems with construct. Because there is no way to edit, modify, or otherwise mangle the inner workings of construct to suit your designs, you have to make everything work as custom behaviors, plugins, or through the event sheet. I like c2, but when it comes to something like this, I would much rather sit down in an environment that is as flexible as possible. Coding collision response in events would suck. Using the sdk isn't as well documented as scripting in unity. Also, unity offers a better environment for that type of development. On the other hand... If you make it using the sdk, you will have constructs editor side speed when you are done. But then you may have to concern yourself over janking in NW.
If contstruct 2 offered an IDE and everything was component based, and physics was better integrated as part of the engine... I would choose construct 2 100% of the time. But right now I flop back and forth between the two.
The major letdown of construct is that it lacks true ObjectOrientedness which is the major problem with it. Using Unity, you can blast apart any problem needing inheritance, scalability, or virtual functions and the like... if you can code. In construct, since families can't inherit from families, and functions are kind of awkward, you lose the speed advantages.
The best example I have is a game Framework I worked on in which I had dozens of characters who could either be controlled by an ai or player. The characters were all as different as in smashbrothers (and as equally the same) and required multiple FSMs. Using families, and passing around UIDs in functions at a high level of abstraction makes c2 lose its charm. It took me two weeks get things to start working, but I had to sacrifice alot of future flexibility. Unity, on the other hand took much less time, as I could simply write abstract scripts how I wanted.
For small things, I use c2. For convoluted projects, I use Unity. I hope this changes in c3, which it may.