Change the mixing logic for polarized and spinor calculations

Description

Change the mixing logic for polarized and spinor calculations.

The code used to mix the spin density matrix. Following Johnson, this is now change to allow for mixing instead the charge density and the magnetization. Moreover, this latter one can now be mixed with a different weight, taken as 1.5 based on the paper of Johnson.

New variables are introduce to test this.

Tests indicate an improved convergence for linear mixing and for Broyden using this new stategy.

Test systems

System	Theory level	Pseudo	Spin	Spacing	K-point grid
bulk iron	LDA	Pseudodojo	spinors	0.3 a.u.	2x2x2
Xe_3^+	LDA+SOC	HGH	spinors	0.42 A	isolated
N-doped graphere nanoflake	LDA	standard	polarized	0.2 A	isolated
Cr_3	LDA+SOC	HGH	spinors	0.2 a.u.	isolated
NiO	GGA+U	HSCV	polarized	0.35a.u.	4x4x4
TiO_2	GGA	from APE	polarized	0.3a.u.	3x3x3
polyFePc	GGA	HSCV	polarized	0.45a.u.	1x1
Li_2Ir_3O_4	LDA+U	HGH	spinors	0.35 a.u.	2x2x2

In all cases, this was tested with

Eigensolver = chebyshev_filter
ConvRelDens = 1e-9
OptimizeChebyshevFilterDegree = no
ExtraStates = 20

News snippet

Change the mixing logic for polarized and spinor calculations.

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.