Done:
- dynamically allocate scheduler context
- see if preferring critical path can improve code efficiency (YES !)

Pending:
- see if dynamically adjusting the critical path leads to further improvements
- test IF
- run result comparison against full set of patches
- check if result comparison actually compares meaningful data
- compare run time and code size for all patches
- see what optimization changes (may interfere with profiling)
- build into Flickernoise (some things may need adapting, e.g., abort())
- review code, see if things can be simplified
- see if valgrind can do something useful