Improve container documentation for foreach and range-based for loop

I have run the attached benchmark to compare the performance of foreach and range-based for using QVector and QHash with clang and gcc for varying container sizes. Example results for container sizes 5, 5000 and 5'000'000 are also attached.

I think the results suggest that Qt documentation on containers should discuss foreach and range-based loop similar to how different threading solutions are discussed at considerable detail. (http://qt-project.org/doc/qt-5/topics-core.html#threading)

According to the benchmark results:

• const QVector in range-for performs significantly better than non-const QVector since the latter detaches when begin() is called.
• Range-for is about 4x faster than foreach for QVector and for all container sizes, i.e. 5, 5000 and 5'000'000
• Range-for and foreach performs similarly for QHash except for container size of 5000000 where detaching non-const QHash takes 3x to 4x longer than the other 3 configurations.