Either I am going mad or Test 1 is broken... (It's quite possibly the former!) Can anyone help?

It's probably a roundoff error in the calculation of the angle (in degrees) by acosd, which causes the end-result to have a slight offset (in the 7th or 8th decimal, which is not printed on-screen but which is there) from the exact correct answer. So instead, use the identity "(sin(x))^2 + (cos(x))^2 = 1" in order to eliminate the use of sin and cos in your calculation.

Lucinda King, you make a fair point: preferably the Test Suite should have allowed for these "slight offsets" that are created when working with floating-point numbers. Especially when the "correct" answers are not integers.   [See e.g. Problem 44690.]

Why is 2.5000 not considered equal to 2.5?