Points for optimization

[Plokji]

• [x] P1. Only decrement semaphore when pop() fails, increases availability to the semaphore
• [ ] P2. distribute sub trees to each of the workers to work on, and then compare values. Normal BnB until stack size > cores, then distribute.
• [ ] S3. implement new bottom level estimation algorithm with an array of bls, or just use completed tasks and bl array
• [ ] P4. google to try find better ways to use multiple threads to traverse a tree.
• [x] P5. Try get the individual workers to periodically compare their best schedule, maybe the NOTIFY keyword will help
• [x] S6. add estimation to phantom node

[Plokji]

• [ ] S3.1 only recalculate if the completed task had the max bl

[Plokji]

• [x] P7. change task to task(*), might help with how paratask load balances.

*done, code is still there, but not sure if it helps at all. Left it in because can't be sure at the moment that the normal TASK calls are limiting the number of cores.

[Plokji]

Scheck if we can amortize data calculation times like how idle time is updated every child