Fix linear solver for metals

[gsavva]

The PR on SIRIUS side should be merged first.
The sirius_linear_solver takes one more argument

[gsavva]

cscs-ci run default

[gsavva]

the reference U value was obtained from native QE HP v.7.2

[gsavva]

strange failure of test LiCoO2 because the scf step did not converge in 100 steps ...

[gsavva]

cscs-ci run default

[toxa81]

@gsavva Can I merge this RP and you continue to debug CI/CD in another PR? Otherwise ci/cd of sirius breaks due to API chage.

[gsavva]

@gsavva Can I merge this RP and you continue to debug CI/CD in another PR? Otherwise ci/cd of sirius breaks due to API chage.

@toxa81 of course, good to go. I've tested the LR fix locally for many metallic systems and the agreement between native QE HP and q-e-sirius HP is excellent.