After all this back-and-forth, nobody disputes the ultimate practical limit to judicial overreach is noncompliance by the other powers, the civil service and even the population. It is sometimes legitimate to refuse to comply to the courts, as is the case for any other part of the State.
But there are different ways to model this in legal theory. The conventional answer would have it that the noncomplied ruling was legal until the very moment the court rolls back its decision. Legality comes from authority, any other interpretation of a law is just an opinion.
Lawyers have their reasons for believing this but there's a radically different proposal if you think of laws in their political context.

Here's a thesis: the point of laws is to call out authorities for not following them. Lawyers will say power comes from laws, but experience shows power has an independent existence and laws as civil law societies understand them didn't really exist in the Ancién Régime. French revolutionaries were the ones who came up with the idea of forcing judges to abide by a single, coherent code, applied equally to all citizens. Judges always had authority. What the law does is force them to justify their rulings to society by demonstrating they comply with a previous written commitment by the State. And if the audience is society, then it is society that interprets what the law means and whether the State is abiding to it. The law is a citizen's shield against the State.
Which answers my first questions: yes, a constitutional court can be wrong about the constitution.