Do you have a clear personal preference for which courses you’d find most engaging independent of how they’ll look on your college application? Maybe you said, but I don’t see it in your first post.
I wouldn’t presume to give advice on college admissions. You probably have a better idea than I do. I can understand the point that AP Physics C is easily identified as a rigorous course with a standard curriculum whereas Web Development is more of an unknown.
But are you excited about learning more physics? If so, that’s a great reason to take it. If it’s mainly to demonstrate rigor, then that may get you above a threshold of consideration, but it won’t set you apart from every other top student who took it to prove they are a top student, and I am sure there are many such people.
Web Development? Who knows? I can see it being taught as a cookie-cutter project, and even then I wouldn’t dismiss it as unworthy. You just have to look at these things realistically.
I think I probably have a better idea of what grad schools look for, because that’s a case where you are expected to show promise in one area and specific research interests. A CS student entering from high school with a current interest in machine learning is pretty likely to shift their interests later. You’re going to get the same fundamentals in any program, including data structures, algorithms, and definitely programming courses. Your interest within CS is unlikely to be relevant for a couple of years.
I would add almost none of the math in physics is useful in even the most mathematical computer science. Occasionally, it can come up. Someone writing a video game engine should have a good grasp of physics (but less so someone programming for an existing engine). There are algorithms like the Fast Multipole Method that involve physics, but I don’t think they’re really standard. The bread and butter of theoretical CS consists of discrete systems, a lot of graph theory, and heavy use of inductive proof.
It’s a little hard for me to put myself in the mindset of someone evaluating a prospective CS undergrad who ultimately wants to pursue machine learning, but that may be my lack of experience. I still think you will need to get through the fundamentals first. If I were trying to predict your success in a top CS department, I would be more interested in seeing your preparation for that (which would probably not be web development, statistics, or calculus, but the math that is rarely taught at all in high school such as induction and logic).