run now forks the benchmark command instead of simply calling exec.
Ressource usage of the benchmark is retrieved with a call to wait4.
But the reported value ru_maxrss differs from values in /proc/<pid>/status.

Calling the loader fails on non ELF executables. Exec, formerly run_cmd,
now takes two options: -l LD_LIBRARY_PATH, -p LD_PRELOAD.
-p sets LD_PRELOAD and -l LD_LIBRARY_PATH before executing the rest of argv.

glibc no longer uses cmd_prefix in favor of LD_LIBRARY_PATH.

Originally the structure of the executed cmd was
{measure cmd} {allocator cmd prefix} {cmd} with the parent environment
except LD_PRELOAD was modified for the whole command chain.
Unfortunatly perf causes segfaults with some allocators and measuring
allocators cmd prefixes doesnt seem fair. So the new cmd chain looks like:
{allocator cmd prefix} {measure cmd} run_cmd <LD_PRELOAD> {cmd} without
touching the environment in python.

run_cmd sets LD_PRELOAD to the value it received in argv[1] and executes
argv[2] with the rest of argv. This does also measure code not part of
the actual benchmark but in a equal manner and not only for some allocators.

Return always 16 Byte aligned blocks and mmap the whole available
memory reported by free -t per thread.

Use systemwide installed allocators by default

Make scans the project directory for Makefiles and executes them.
The targets should be build by the new targets/Makefile.
All included Makefiles are now "quiet" by default.

each benchmark has its own Makefile which must put it's binaries into
OBJDIR which is added to the PATH during execution.