Missing tutorial: non-linear problem with strongly enforced Dirichlet conditions

[ordinary-slim]

There is no tutorial where the multiphenicsx variants of set_bc and apply_lifting are used with a non-default x0 argument (default is None). An example with a non-linear problem and strongly enforced Dirichlet conditions could feature these.

[francesco-ballarin]

Also nest cases are currently missing from tutorials. See #39

[francesco-ballarin]

Plan of action after #47 is merged:

1. add a new tutorial directory with two cases:
   1. domain decomposition with overlap (Schwarz alternating method)
   2. domain decomposition without overlap (Dirichlet-Neumann scheme and Robin-Robin scheme inspired by Robin-Robin DD example #39)
2. parametrize tutorials on kind. In Updates for upstream PETSc assembly changes #47 kind="mpi", will be hardcoded, but at a later stage it will be replaced by a parametrization between kind="mpi" or kind="nest". CI will switch from one to the other via an environment variable, with default value "mpi".
3. After this, double check if
   "An example with a non-linear problem and strongly enforced Dirichlet conditions could feature these."
   is still needed. Most of the tutorials will now show the high-level interface to PETSc solvers, introduced in Updates for upstream PETSc assembly changes #47. There are still a few places where the low level interface is shown: some of those tutorials will stay like that, others may move to the new high-level interface.
4. I can't remember the use case for Change set_bc and set_bc_nest to allow either restricted or unrestricted x0 #34. If there was a simple problem @ordinary-slim could explain to me, I could consider determining if we need a tutorial for that or not.

[ordinary-slim]

1. add a new tutorial directory with two cases:
   
   1. domain decomposition with overlap (Schwarz alternating method)
   2. domain decomposition without overlap (Dirichlet-Neumann scheme and Robin-Robin scheme inspired by [Robin-Robin DD example #39](https://github.com/multiphenics/multiphenicsx/discussions/39))

Right now all the present examples are for monolithic DD, so Schwarz alternating example is welcome. I can contribute staggered non overlapping DD examples (Dirichlet Neumann or Robin Robin), let me know if that is of interest.

   "An example with a non-linear problem and strongly enforced Dirichlet conditions could feature these."
   is still needed. Most of the tutorials will now show the high-level interface to PETSc solvers, introduced in [Updates for upstream PETSc assembly changes #47](https://github.com/multiphenics/multiphenicsx/pull/47). There are still a few places where the low level interface is shown: some of those tutorials will stay like that, others may move to the new high-level interface.

I learned "low level" interaction with PETSc through the multiphenicsx tutorials, I am glad they were there.

4. I can't remember the use case for [Change `set_bc` and `set_bc_nest` to allow either restricted or unrestricted `x0` #34](https://github.com/multiphenics/multiphenicsx/pull/34). If there was a simple problem [@ordinary-slim](https://github.com/ordinary-slim) could explain to me, I could consider determining if we need a tutorial for that or not.

It comes up naturally when solving a non-linear problem with a Dirichlet BC. Before the PR, the x0 had to be unrestricted, which seemed unintuitive to me when working with restrictions, see https://fenicsproject.discourse.group/t/issue-with-multiphenicsx-fem-petsc-set-bc-x0-argument/16089.