Ability to re-execute benchmarks

[PhilippWendler]

BenchExec could get an additional benchmarking mode, where it does not get a benchmark XML file, but a result file, and then re-executes all the runs that are present in the result file.

Tasks:

• [ ] check whether all necessary information is present in the result file
• [ ] implement feature
• [ ] compute difference in experimental setup and warn user (e.g., if machine model, tool version, environment variables, etc. differ)