And fix all spelling errors that codespell currently finds.

See https://reuse.software/ for a description.

Nested sparse matrices were not allowed to have empty rows/cols. By choosing the minimum offset of other blocks, we get an educated guess for these cases.

Simplify the code while we're there.

Is not used.

Simplify the implementation while we're there.  The IsNumber trait
allows much simpler recursion termination.

The code still involves recursion, therefore the separate 'Helper'
method remains necessary.

Previously, there was a list of special matrix types that ended
the recursion (FieldMatrix, DiagonalMatrix, etc.).  Simply
ending the recursion at number types simplifies the code
considerably, and allows to write matrices like BCRSMatrix<double>
and such.

This allows to call writeMatrixToMatlabHelper for matrix types
that do not actually implement true matrix rows, such as
DiagonalMatrix and ScaledIdentityMatrix.

Using `unsigned long` instead of `size_t` only works on architectures
where the two types are identical, but not on, for example, i386.  On
architectures where the types differ, the generic version would be
used instead of the specialization for `std::array`.

Bug-Debian: https://bugs.debian.org/860666

with g++ 6.3 the  specialization for std::array failed, as the integer type
was unsigned int instead of std::size_t

similar to Matlab matrix writer

[[Imported from SVN: r1730]]

[[Imported from SVN: r1713]]

ScaledIdMatrix or a DiagonalMatrix.

Patch by Uli Sack

[[Imported from SVN: r1636]]

[[Imported from SVN: r1600]]

(Aplied patch of Matthias Wohlmut with a small modification that
resets the precision to the old value).
Thanks go to Matthias.

[[Imported from SVN: r1508]]

file in the matrix market exchange format.

[[Imported from SVN: r1488]]

[[Imported from SVN: r1341]]

[[Imported from SVN: r1284]]

[[Imported from SVN: r1282]]

[[Imported from SVN: r1262]]

[[Imported from SVN: r1210]]

[[Imported from SVN: r1207]]

[[Imported from SVN: r1206]]