Hi Steve. Just read the book in one sitting. Fantastic stuff and great to see all you've made available through production tested code!
I was a little surprised at the light treatment on parallelism -- on the one hand, side-effect free functions practically beg to be run as threads/tasks -- on the other, many libraries including RxCpp shy away from choosing among the many possible dependencies, perhaps waiting for a standard to emerge.
You seem to leave it to the implementor to decide how to parallelize their stream and cell functions -- in practice, has that flexibility outweighed the potential complexity it adds to stream and cell function definitions?