python typehints: reduce the number of runtime-type-check wrappers

The --runtime_type_check flag is considered to be a performance penalty.
One of the possible slowdowns is the wrapper around DoFn.process(), which adds an extra call to the wrapped process() method.
Currently, there are 2 wrappers applied: OutputCheckWrapperDoFn and TypeCheckWrapperDoFn. These could be merged into one to possibly save on overhead.