This is not how the Categorical Imperative is conceived. The CI is a legalistic statement of this type: "I will do (or allow) X if and only if I can reasonably will that that anyone can do (or allow) X". The intention is to force the person contemplating an act to consider the wills of others: might persons A, B, and C also will this to be the case? The question isn't whether everyone should take their own life; the question is whether anyone — you, your parents or children, the person you hate most, a random stranger on a bus — should be allowed to take their own lives. If we say to ourselves "it's ok for us or someone we hate to do it, but it is not ok for our children to" then we cannot will it as a universal law, and thus we should not do it.
The main weakness of the CI as moral theory is that it tends to collapse into solipsism. To use your example, yes: The Marquis de Sade enjoys both giving and receiving pain, and so he could certainly hold it as a universal principle that everyone should have the right to inflict torment. But what Kant wants to avoid is allowing the MdS to impose his will on others against their will. MdS would have to craft his universal law as something like "anyone may do it to anyone if they both will it to be so" to have any hope of satisfying the CI. One cannot torment someone out of boredom unless the other wills that torment to be so.