Allows for subspace diagonalization of nonorthonormal states.

Description

Allows for subspace diagonalization of nonorthonormal states. This is used to speed-up the Chebyshev filtering.

The calculation of the overlap matrix and Hamiltonian matrix are combined in order to reduce the number of communications and memory transfer.

Removes one unnecessary subspace diagonalization for the first step of the Chebyshev filtering.

News snippet

Improvements for the Chebyshev filtering

Checklist

• I have checked that my code follows the Octopus coding standards
• I have added tests for all the new features added in this request.