Because equal does not use all the techniques available to Maxima for
equality checking.
Why not? Because a) some of the "stronger" techniques give incorrect
results in some cases and b) some of the "stronger" techniques can
take very large amounts of runtime in some cases.
For this particular example, you could try defining
rcequal(a,b) := is(radcan(a-b)=0)
But beware -- radcan is sometimes sloppy (by design) about the signs
of radicals.
-s