- Replaced Constants.h with std::numeric_limits, see also
  https://jlk.fjfi.cvut.cz/gitlab/mmg/tnl-dev/issues/3
- Made the initialValue method of reduction opeerations static and
  constexpr.

* cineca/mpi:
  Added specialization of MPIDataType for bool
  Added missing operations to MpiDefs.h
  Added assertions to MpiCommunicator.h checking that MPI functions are not called with the null group
  Whitespace and typo fixes in MpiCommunicator
  Fixed the TNL_MPI_PRINT macro
  Fixed signature of the Allreduce method of communicators
  Added NullGroup to communicators because it is useful for initializations
  Removed empty file and directory in UnitTests

Added assertions to MpiCommunicator.h checking that MPI functions are not called with the null group

Decoupling linear solvers from PDE solvers

See merge request mmg/tnl-dev!5

Also added missing 'const' to the getMemoryElement method for host.

[skip ci]

Refactoring: replaced TNL::SharedPointer with std::shared_ptr for the linear solver and preconditioner

The DeviceType capability does not matter because these objects are
never passed to the device and we need to pass "shared_ptr<Derived>" as
"shared_ptr<Base>".

They are virtual, currently not needed and only complicate things.

The vector view types have to be passed by value, because pass by
reference does not consider the converting constructors.

The x and b vectors might have different type.

It is actually used only in 2 methods and unnecessarily complicates the
interface for all other solvers.