I am asking for a set among n competing hypotheses where the simplest one was H_i but the correct one was H_j where i is not the same as j. That is, a time when scientists tested a bunch of hypotheses and found that the more complex one was correct.
The situation you describe would not be a counter-example to Occam's razor but just a reminder that it's a rule of thumb. It does not say that the simplest theory is always correct, it says that given two equally effective solutions the one that requires the least hypotheses or hypothetical entities is to be preferred.
It is very common for a theory to turn out to be be too simple to work and to require extra theoretical entities to make it workable. Galactic rotation might be a good example.