There is a lot to say about this... You can find more details in the literature, however, the Krylov-Newton can be better when you have materials with strain softening. While I found the newton line search better for contact problems.What is the difference between Krylove-Newton and Newton with Line Search?
Yes, in fact you solved the problem with this.Does it make sense for me to change this?
Now it's fine, from the image I can see that your analysis finished successfully.Is there any other way to converge the model?
Probably you will have other issues if you will start using nonlinear material for soils.
But nonlinear analyses are like this... You have to try different options, there is no unique rule...