First off, I want to say this is a good question. Second, I agree with your objection -- we cannot appeal to "good" (especially without definition at this point).
The issue you are highlighting is one that is well known to Kant scholars (here I mean on 2nd critique / ethicists). Kant's claim that we can test morality with universals sounds like a great idea and one that would get us a universal, objective morality founded on reason alone. The devil is in the details.
What you call cuteness is basically one of those devils. Or to put it another way, there's a critique that articulates the same issue this way: somehow the things Kant cannot universalize are those things that are incompatible with bourgeois Prussian society centered around the town of Konigsberg.
Here, the theory:
(1) For any maxim M, we can test whether the action is good by seeing whether we could simultaneously will that all agents in situation S should will the same.
(2) "good" here means compatible with objective reason, i.e. universalizable. I can get the reference if you want. But specifically this is *objective good* which Kant distinguishes from things we like to experience. (If my memory is correct, he does not use the opposite phrase subjective good).
Now this definition actually does some of the legwork for us. But there's more that is helpful in Kant.
Returning to the two examples from the text (which based on your choices I'm going to assume is the Groundwork), we can see why Kant still (probably) can say they are wrong:
In the case of lying, it's not merely that truth is lost but that the ability for other agents to act rationally is lost. Thus, it impugns the rationality of creatures and is thus immoral.
Similarly, property is for Kant a rational feature of things (this is expounded in the Metaphysical Doctrine/Principle of Right). Thus, to undo it is to remove something he takes to be necessary to rationality. (This one is clearly the more dubious of the two proofs since the definition of property is far from evident in our retrospective capacity).
Also, I must add as an addendum that there are two variants of the universalizability test in the Groundwork. One checks whether it is incoherent on a purely logical basis. The other asks whether it could be a universal law of nature. On my interpretation the latter adds what one takes to be the necessary laws of the universe.
Moving to three your suggested tests for universalization.
Drugs for Kant will need to get treated in the same way as alcohol as Mozibur notes. Kant has such a treatment in the Metaphysical Principles of Virtue for alcohol. Working from memory, drinking is okay, but being so drunk that you are not rational is not. Thus, the same question is going to arise with drugs. If you become irrational, it is immoral for Kant, because morality is about acting rationally. And whenever a creature capable of acting rationally doesn't, that is immoral.
For the not giving to the poor, that's not actually a duty for Kant in the strict (perfect) sense. So he would agree that there's no logical contradiction. It is a form of imperfect duty under the obligation to help others. But this duty, at least as far as I argue, is one we hold insofar as we creatures who sometimes need help. This is already in the Groundwork but gets repeated with some details in the Metaphysical Principle of Virtue
Smoking for Kant would be fine if it does not impugn rationality or would need to be treated in the same way as obesity (also handled in the Metaphysical Principles of Virtue
What might interest you is the treatment by Allen Wood in Kant's Ethical Thought that looks at some more interesting objections that occur in terms of even defining the maxim to be universalized. Or I might be getting confused with this text by Henry Allison. But there's several questions as to what a maxim even is.