* where \c <integral \c type> stands for any integer type compatible with Eigen::Index (i.e. \c std::ptrdiff_t).
*
* The last statement implies compatibility with \c std::vector, \c std::valarray, \c std::array, many of the Range-v3's ranges, etc.
*
* If the submatrix can be represented using a starting position \c (i,j) and positive sizes \c (rows,columns), then this
* method will returns a Block object after extraction of the relevant information from the passed arguments. This is the case
* when all arguments are either:
* - An integer
* - Eigen::all
* - An ArithmeticSequence with compile-time increment strictly equal to 1, as returned by Eigen::seq(a,b), and Eigen::seqN(a,N).
*
* Otherwise a more general IndexedView<Derived,RowIndices',ColIndices'> object will be returned, after conversion of the inputs
* to more suitable types \c RowIndices' and \c ColIndices'.
*
* For 1D vectors and arrays, you better use the operator()(const Indices&) overload, which behave the same way but taking a single parameter.
*
* See also this <a href="https://stackoverflow.com/questions/46110917/eigen-replicate-items-along-one-dimension-without-useless-allocations">question</a> and its answer for an example of how to duplicate coefficients.
*
* \sa operator()(const Indices&), class Block, class IndexedView, DenseBase::block(Index,Index,Index,Index)