Experiment with DejaFu or IOSim

[NicolasT]

Currently, Process r a is defined as ReaderT (ProcessEnv r) IO a. Indeed, we don't want any other monad but IO as the base monad for processes (theoretically anything implementing MonadUnliftIO may work, dunno, but it's not really worth it since that's then limited to ReaderT and IdentityT anyway).

Now, it could make sense to have the internal implementation to be newtype ProcessT r m a = ProcessT (ReaderT (ProcessEnv r) m) a, and only export type Process r = ProcessT r IO in the public-facing API (and specializing all functions for it, as is already the case). Then, the implementation could be abstracted over MonadConc (or similar) and we could write tests using DejaFu or IOSim to validate certain behaviours.

Lots of opportunity for experimentation here.