vector should be for mathematical vectors
This is the only sense of "vector" that isn't more clearly expressed in a different way. There might not yet be any relevant questions for that concept, but I don't see a reason to outlaw it preemptively. I can easily imagine someone wanting to design a programming language that has native support for a data type that behaves like a mathematical vector, not to mention that language's standard library.
simd is good enough for SIMD stuff
Because really, what other things do you use in SIMD besides what it calls "vectors"?
dynamic-array is the clearest term for the other thing
Other languages/ecosystems refer to this concept as "vector" (as noted) or "list"; but these terms are ambiguous. You already noticed the problem with "vector"; "list" can also mean a linked list (and often does). By contrast, "array" can't realistically mean anything that isn't a sequence of "real" (i.e., not determined on the fly) values that are actually stored sequentially in memory (although they might be indirected), offering random access.
However, "array" by itself carries a strong connotation of pre-sized storage without any book-keeping information for the size (as in C, for example). I propose to use array to mean such an array of a fixed size; and dynamic-array to mean what Python calls a list, C++ calls a vector, etc.