Make solvers more uniform

[MaxenceGollier]

A few points that need to be adressed

• Use implicit updating of the different linear operators instead of recreating one at each iteration
• Solve the arrow issue when verbose > 0 across solvers Use callback docstring for all solvers and solve the bug with the arrow being printed too far with verbose #234
• Use the callback variable for the doc: Use callback docstring for all solvers and solve the bug with the arrow being printed too far with verbose #234
• Replace \nu in R2 with \sigma k (like R2N and R2DH) : R2: remove nu keyword argument and replace with sigma #232
• Make it so that solvers do n iterations when max_iter = n instead of n+1, also move get_status from src/R2_alg.jl to src/utils.jl Move and update get_status #233
• Solve the issue with the way the step length is computed in LM/LMTR/TR