This doesn't have a simple answer.
Problem 1 - your calculation is for a perfect, ideal metal. The actual calculation (first done by Lifschitz) for a physical material is much messier.
In 1968, Boyer calculated the Casimir effect for a bubble of vacuum in a sea of metal stretching to the ends of the universe. He found this to be repulsive and infinite. This is probably the "classic" calculation, but while not exactly wrong, it's not exactly relevant. In the rectangular Casimir effect, the finite force is obtained by subtracting the contribution from the inside from the contribution from the outside (or vice versa). In Boyer's configuration, there is no outside.
Additionally, Boyer assumed that his sphere only radiates in directions normal to its surface. If you were to look at such a sphere, you would only see one small point - the point that happens to be directly in line of sight. A proper calculation for a physical material needs to consider emission in all directions. I don't know if such a proper calculation has been done - if the Lifschitz calculation is messier, this is messiern squared, but the force would be attractive. (Replace the sphere with an 2n-hedron of parallel plates - the force between all n opposite plates is attractive, so therefore the total force is attractive. Now let n go to infinity)
This is for a static sphere. For a collapsing sphere, it will be even more complicated.