It is not needed, since CUSPARSE is included in all versions of CUDA
since at least 7.0.

Also note that CUSPARSE is needed only for some tests and legacy code
(tnl-benchmark-spmv and wrappers in Matrices::CSR) and experimental code
(Preconditioners::ILU0). Everything is still guarded by the
HAVE_CUSPARSE macro, so the user can pass -DHAVE_CUSPARSE -lcusparse to
the compiler to enable the features.

This way they can use either yes/no, on/off, or true/false.

Implemented Array:containsOnlyValue.
Restored implementation of ArrayOperations::memoryCompare on host.

Implemented a method Array::checkValue which returns true only if there is an element with given value in the array.

I don't understand how it could work on the small grids...

The Host class was combining too many things...

It actually does not affect anything...

Why do problems have this method anyway? It was obviously never used...

When type traits for the "double-float" or "double-double" types are
needed, they should be added as specializations of the
std::numeric_limits template in some of the TNL header files.

There were 2 problems:

- it does not belong into a numerical library
- the approach with hardcoded escape codes is not portable even on Linux

There are separate libraries which solve both of these issues much more
nicely, they should be used instead.