chore: Refactor linear solver to have pre conditioner baked in the same class

[anushkasinghh]

Refactor linear solver to have pre conditioner baked in the same class. Pre conditioner is now a parameter rather than a different class altogether.

The following is done in the pull request more granularly :

1. Merge ConjugateGradient and PConjugatreGradient
2. Merge BiConjugateGradientStabilized and PBiConjugateGradientStabilized

Will solve #469 when merged

[anushkasinghh]

While debugging the bicgstab solver (using the old implementation of the algorithm), I found that it consistently fails for complex systems when using a diagonal preconditioner (pc = A.diagonal(inverse=True)). However, all tests pass when I set pc = None. This issue appears to affect only bicgstab with complex dtype — all other solver/dtype combinations behave as expected.

It looks like the preconditioner is destabilizing the method for complex matrices. Would you recommend skipping the preconditioner in this specific case @yguclu ?

========================================================================== short test summary info ===========================================================================
FAILED psydac/linalg/tests/test_solvers.py::test_solver_tridiagonal[bicgstab-complex-2-5] - assert np.float64(1.8472203713614275e+24) < 1e-08
FAILED psydac/linalg/tests/test_solvers.py::test_solver_tridiagonal[bicgstab-complex-2-10] - assert np.float64(786.0899812535919) < 1e-08
FAILED psydac/linalg/tests/test_solvers.py::test_solver_tridiagonal[bicgstab-complex-2-13] - assert np.float64(8.004021243308681) < 1e-08
FAILED psydac/linalg/tests/test_solvers.py::test_solver_tridiagonal[bicgstab-complex-3-5] - assert np.float64(49927742822871.3) < 1e-08
FAILED psydac/linalg/tests/test_solvers.py::test_solver_tridiagonal[bicgstab-complex-3-10] - assert np.float64(0.0005220081883679691) < 1e-08
FAILED psydac/linalg/tests/test_solvers.py::test_solver_tridiagonal[bicgstab-complex-3-13] - assert False
================================================================== 6 failed, 66 passed, 5 warnings in 4.79s 

[anushkasinghh]

changes migrated to pr512