v0.8.0

Update to use parallel MIP via HiGHS v1.15.

This update provides a significant boost in performance across the whole range of optimized budgets.

It can, however, also be significantly slower on particular instances. With the old setup the solver was run using parallel processes that would race to be the first to finish and each worker process started with a different random seed then the workers would attempt to share the best solutions as they found them. The new built-in parallel setup uses a single random seed and the workers are assigned portions of the problem to work on in a deterministic manner. That means that if a 'bad-seed' is used that is just luck of the draw -vs- the previous setup where you'd need to have 'n' bad seeds at the same time to have a poorly performing solve instance.

Since the overall gains are significant (30-50% improvement) and we are all used to BDO-rng it seems a good trade-off to me.