I explain at the start of the first test that, "Questions clarifying what you're reading are okay (raise your hand, I'll come over). I cannot tell you how to solve a problem. My response may be, 'I can't tell you that during the test.'"
If Question #1 comes up, I do as stated. Questions #2-3 get a fairly loud but(but polite) repetition of, "I can't tell you that during the test." (Or, "That's what you're being tested on."). Question #4 may get, "Read"Just read the direction carefully" or sometimes actual pointing at it(which itself is a repetition of something I need to say during almost every single class meeting).
I find this public act tamps down on questions as much as I can hope for. The main thing is to be solidly consistent. Counterexample: I was co-proctoring with another professor in their class. That professor got a question, shook his head in disgust, muttered "You should know that", and walked away. Then turned around, walked back, and actually did walktalk the student through how to solve the problem. This can only encourage more fishing questions like that.
Due diligence: In liberal-arts math courses I have occasionally had students express disbelief and/or outrage that I wouldn't help them during a test. "But you're the teacher, that's your job" kind of thing.