Benchmarker: Legend in pictures are too big => pictures are not readable

I just ran a benchmark in which I wanted to test a few possibilites :

• 1 dataset (in the final test I wanted to test 4, but started with only 1)
• 5 possibilities for Local Search
• 3 configs for Construction Heuristic
• 3 possible durations to run the solver
• the possibility to run or not a second phase of Construction Heuristic+LocalSearch

This does not seems an enormous benchmark to me but it already results in 90 tests...

The unfortunate consequence of this large number of tests is that the pictures presented in the benchmark report are unreadable, since the legend covers the whole screen.
See attached picture.
For more readability, maybe the legend could be put out of the picture or only the X best solution presented there?