Does this hold even if we don't include whales and dolphins in "things that look like fish"?

Those aren't the problem. The real issue is that the tetrapods which evolved into most land animals (amphibians, reptiles, birds, mammals) are further down the phylogenetic tree of bony fishes than coelacanths and lungfish, which are further down the tree than cartilaginous fishes like sharks and rays, which are further down the tree than jawless fishes like lampreys and hagfish.

In taxonomy, it's called a "Paraphyletic group" [1].

[1]: https://en.wikipedia.org/wiki/Paraphyly#Examples

Yes, the issue is the ancestry between "fish" being very distant. It doesn't matter if you exclude marine mammals. Many fish in the ocean are still more closely related to beings on land than another fish. It's the equivalent of calling all flying animals birds. If we excluded bats from this new definition of "bird" then a bumblebee won't suddenly become more closely related to a Buzzard.

It surely does. This website is a good way to visualize the common ancestor of the bottlenosed dolphin and zebrafish. It's the same common ancestor as a human and a zebrafish, or a bird and a zebrafish. It's an ancient ancestor!

https://www.onezoom.org/life/@Gnathostomata=278114?otthome=%...