Add a small section for inf or nan.

Closes #129 (closed)

requested review from @vrouvrea

assigned to @sgilles

Should we talk about this behaviour:

    std::cout << "double [min, max] = [" << std::numeric_limits<double>::lowest() << ", "
        << std::numeric_limits<double>::max() << "]" << std::endl;
    // double [min, max] = [-1.79769e+308, 1.79769e+308]

    // You must substract or add something significant enough to be taken into account
    std::cout << "double [min-1.79769e+300, max+1.79769e+300] = [" << std::numeric_limits<double>::lowest()-1.79769e+300 << ", "
        << std::numeric_limits<double>::max()+1.79769e+300 << "]" << std::endl;
    // double [min-1.79769e+300, max+1.79769e+300] = [-inf, inf]

    std::cout << 0. / 0. << std::endl;
    // nan
    std::cout << 0 / 0 << std::endl;
    // Floating point exception(core dumped)

Which is not the same if the type does not have an infinity or nan values

I didn't know so if you can add it it would be nice!

I would mitigate this sentence, maybe ... All types stored especially don't go from minus infinity to infinity Maybe an example like:

    std::cout << "int [min-1, max+1] = [" << std::numeric_limits<int>::lowest() - 1 << ", "
            << std::numeric_limits<int>::max() +1 << "]" << std::endl;
    // int [min-1, max+1] = [2147483647, -2147483648]

To be discussed...

Just below this sentence there is a wall a min/max, and just below the illustration of what happens if we add 1... So for me all is already there, but clearly it's not properly presented. The best is maybe that you propose an alternate way of presenting it? (I don't feel up to the task as I'm fine with current content...)

mentioned in merge request !105 (merged)

closed