To your specific points:
*”Physical Sciences” is definitely an exception, though its typically called (at least in the US) just “Physics”.
* I’d say creating an algorithm is more engineering than science, though it’s a gray area.
* In general, I agree you’re right that x science is an attempt to treat x in a more scientific manner than previously. In that sense, it’s good.
* Aha, good point: Computer sciences, plural, goes well with the interdisciplinary/diverse nature of the field.

By this token the School of Physical Sciences at Cambridge (to take one such random example) would be a bunch of quacks – do you believe that? I think whenever we use the term x science then we are really aspiring to come up with a scientific basis for x, and it is perfectly fine if the methods have their roots in another place but the question is one of focus – if you had to choose, would you prefer to prove one more theorem on asymptotics of something (even it is essentially irrelevant to the subject matter of x) or would you develop a model/algorithm/empirical experiment around the core of x? The former would be the preference in an original area while the latter justifies the spawning and growth of a new area.

BTW, my own preference is for the term “computer sciences” (plural, indicating diversity), as in the name of the department I received my doctorate from.