Fised index type in assertions in getView/getConstView methods in Array/ArrayView/Vector/VectorView.

- move constructors and assignment operators should be noexcept
  https://clang.llvm.org/extra/clang-tidy/checks/performance-noexcept-move-constructor.html
- avoid unnecessary copy on range loop variables
  https://clang.llvm.org/extra/clang-tidy/checks/performance-for-range-copy.html
- avoid unnecessary value parameters of types that are expensive to copy
  https://clang.llvm.org/extra/clang-tidy/checks/performance-unnecessary-value-param.html
- avoid unnecessary copy initialization of types that are expensive to copy
  https://clang.llvm.org/extra/clang-tidy/checks/performance-unnecessary-copy-initialization.html

...and in a few other places to avoid compiler errors/warnings.

Jakub Klinkovský authored 3 years ago and Jakub Klinkovský committed 3 years ago

The old headers were hard to maintain, inconsistent and contained
useless information:

- file name (useless, prone to inconsistencies upon renaming)
- "description" (useless placeholder that was likely intended to be
  replaced by the user, not to be used literally)
- "begin" date for each file (prone to inconsistencies and copy-paste
  errors, the information can be found in git)
- "copyright" notice (inconsistent years)
- "email" (replaced with a project link in the new template)
- "Copyright Notice" referring to an obsolete file (tnl/Copyright) that
  does not exist anymore

The new template consists of three simple parts:

- copyright notice with updatable years related to the whole project
- project name and website URL
- SPDX license identifier

Note that the SPDX license identifier must be on a line that starts with
either "//" or "/*", so we can't use a block comment (/* ... */) for the
whole header. But using "//" for block comments is common too.

Replaced send/receive for Array(,View) and mpiSend/mpiReceive for String with a general implementation in the MPI namespace

Also added analogous functions: MPI::sendrecv, MPI::bcast.

The algorithms are implemented as plain functions in TNL::Algorithms.
containsValue was replaced with contains.

Removed reduction methods from Array and ArrayView, instead added overloads of reduce and reduceWithArgument for arrays/views

Plain functions are much more flexible than methods. The methods were
also violating the open-closed principle:
https://en.wikipedia.org/wiki/Open%E2%80%93closed_principle

Hence, all StaticArray, Array, ArrayView and even expression templates are
directly usable in reduction without the need to create a wrapping fetch
functor. Also NDArray has this interface in 1D.

- passing indices and array/vector values by reference is useless
- declaring arguments passed by value as "const" is useless
- using "IndexType" and "ValueType" is better than "Index" and "Value",
  because Doxygen creates an automatic link to the description of the
  type
- various other documentation and whitespace fixes

- if ValueType is a fundamental type, the Array behaves the same as
  before (allocated elements are initialized to indeterminate values)
- otherwise, allocated elements are value-initialized using
  MemoryOperations::construct and destructed using
  MemoryOperations::destruct before their deallocation

Serialization in TNL::File: File::save and File::load are specialized by Allocator instead of Device

The usage of algorithms such as MemoryOperations or Reduction is not
bound to a particular container. On the other hand, ArrayIO,
ArrayAssignment, VectorAssignment and StaticArrayAssignment are just
implementation details for the containers - moved into
TNL/Containers/detail/

Also moved ParallelFor, StaticFor, StaticVectorFor, TemplateStaticFor
into TNL/Algorithms/

This will be necessary to avoid code bloat with more than 2 devices
(execution types).

Fixes #46