Adds a parallel manager, allowing parallel execution of reaktoro blocks.

[avdudchenko]

Realted to 9 this adds ability for reaktoro to run in parallel using multiprocessing.

This PR added a a new ReaktoroBlockManager class, that can be passed into a ReaktoroBlock allowing its aggregation and solving in parallel.

Most functionality remains the same. Typical usage is shown below :

m.reaktoro_manager = ReaktoroBlockManager()
m.property_block_1 = ReaktoroBlock(reaktoro_block_manager=m.reaktoro_manager, other options...)
m.property_block_2 = ReaktoroBlock(reaktoro_block_manager=m.reaktoro_manager, other options...)
m.property_block_3 = ReaktoroBlock(reaktoro_block_manager=m.reaktoro_manager, other options...)
m.reaktoro_manager.build_reaktoro_blocks()

Proceed to use as normal in flowsheet e.g.

init model 
m.property_block_1.initialize()
m.property_block_2.initialize()
m.property_block_3.initialize()
solve model 

once completed working with flowsheet ensure to kill all workers with
m.reaktoro_manager.terminate_workers()

Left to add:

• async for when there are more blocks than the maximum allowed processors - currently, no checks or limits are enforced.

• Currently, there is no clean way to interact directly with reaktoro_block.reaktoro_model.outputs.

• Fix display for reaktoro state.