Linear combinations of these solutions still satisfy Schrödinger's equation by linearity but they are not stationary because the E differ
except for Hydrogen-(like atoms), atomic orbitals are NEVER a true solution of the SDE. They can't be. The Hamiltonian contains terms which up to two electron coordinates. An orbital, by definition, only contains one electron coordinate.
I would put it like that:
-For an exact treatment, we would need to solve the Schrödinger Equation with the molecular Hamiltonian.
-We can't find an analytical solution.
-For very small Molecules, there are very exact numerical solutions (not analytical doesn't mean not exact!)
-For every Molecule that Organic Chemists are interested in, we would need sophisticated and computer intensive algorithms. Even today, these results don't necessarily give exact solutions.
-We CAN get an intuitive understanding with qualitative MO-Theory (and or Valence Bond, however I still don't see the qualitative difference between them...)
-For that, we build the Molecular Orbitals from Atomic Orbitals (LCAO)
-If we have a saturated carbon atom, we could, in prinicple, use the atomic s and p orbitals, it's just very inconvenient. That's why we combine them BEFORE we actually build combinations with other atomic orbitals in the Molecule. That's all there is to Hybridisation. It's nothing spooky.
The original paper about hybridization could be clearer than what has been written later - it's often the case.
Do you know "the" original paper? The term definitely seems to be historically overused.