mdtest computes wrongly with stonewall

[JulianKunkel]

The computation of mdtest when using stonewall was wrong. It used the number of items instead of the actual achieved stonewall number!

To verify if your mdtest version is still affected, run:
$ mdtest -C -W 1 -u -L -F -n 100000000
This will run mdtest for 1 second to create files, you can check if the reported value is too high as it will output as well:
"Continue stonewall hit min: X"
Then you should see that mdtest reports for File creation approximately X, too. Not 100000000.

Sorry about this. Have to track where the issue has been introduced.

[JulianKunkel]

The bug is fixed with 4297ccd

[JulianKunkel]

Started to work on a testing platform to nail down such matters in the future and cover more aspects.
See the branch here

Due to the complexity of MDtest there may be some branches when users use sets of flags are used that are affected by this as there are many possible routes through the code. the wallclock feature was initially implemented to first suit the needs of IO-500 but will be expanded and corrected for more cases.