Fine-grained locking not feasible due to time constraints

Calculating lock statutus per handle and op takes too long time to be feasible.  It would give a better picture of how memory is actually used, but unfortunately it cannot be calculated in something close to realtime - even pushing it out to many workers on a fast CPU takes well over two days and yields data proportional to the size of the number of ops.

Coming up with a better way to calculate "micro" lifetime is something for the future work-section.