This can be used to run slow or blocking operations on the secondary
pool without having to start an entirely separate process. While the
latter might feel more natural it requires more memory (at least 32 KB
the moment this new process allocates an object) and requires copying of
data sent between these processes. On the other hand moving a process
from one pool to another is cheap and doesn't suffer from any of these
issues.

Using this mechanism one could implement a file reading method like so:

    def read -> String {
      process.run_in_secondary_pool {
        _INKOC.file_read(@File)
      }
    }

With this change the method `process.spawn_io` is removed as it most
likely won't be used for a while (if ever).