(A bit more than an answer; too long to be a comment. Slightly opinionated.)
Yes, top journals do this. That said, publishing in other journals is ok. And sometimes an idea seems obvious to one person only seems obvious because they are now reading someone discussing it or proving it. And even when yes it would be obvious to some people, that's still ok.
It is genuinely helpful to have in the literature papers that might be easy or obvious to some people. One shouldn't think that because something isn't going to be published in a top journal that there's anything wrong with it. That doesn't make it not good mathematics.
In that context there are two types of papers which are much helpful enough to be worth stating explicitly as yes-that's-good-math and which shouldn't be published in top journals and that's ok. The first is papers which are primarily computational where there's at most only minor improvements in algorithms used, but where one is taking advantage of better computing power. That said, when one does publish those sort of papers, an additional goal should be to make them have at least better expository results. The second is papers which take a bound using some asymptotic estimate like the prime number theorem and then either make that bound explicit or use improved versions of known bounds. For example, there are a lot of papers which use Rosser and Schoenfeld's explicit bounds on the PNT to get explicit bounds on some other object of interest. Since we now have a whole bunch of bounds which are tighter than Rosser and Schoenfeld, using them to tighten up the results which depended on Rosser and Schoenfeld can be helpful. But it is obviously not the sort of thing which should go in a top tier journal.
I'd really be happy with both of these sorts of papers getting their own explicit journal. Something like "The Journal of Straightforward Improvements."