[wsrp-interfaces] Layering of spec and joint spec

My belief is that we need both (1) a layered spec -- to support
extensibility -- and (2) a fairly broad single interface at the base layer
-- for good coverage in the marketplace of commonly expected use cases.

In particular...

1. Once we have factored the interface along dimensions such as those
outlined by Mike (with/without entities, stateful/stateless, properties/no
properties) we can identify a simple set of operations that pertain to
lifecycle, output, and actions.

2. I believe these core operations should be part of a single base
interface layer.  Generalizing these into the language used in the joint
spec is as much to benefit WSRP as to support interoperability with WSIA
since some of the original WSRP use cases include more than portals as
Consumers and Producers.

3. Many other WSIA features are likely to be core to layer releases of
WSRP.  When dealing with coordinating runtime state across portlets, or
dynamically configuring portlets created at runtime, the session-level
property interface now being considered by WSIA will over time likely
evolve to be a core WSRP requirement as well.

For all of these reasons, I believe continuing the joint work between WSRP
and WSIA interfaces is a good idea.  There will nonetheless be packages of
the interfaces that will be of more interest to one market segment than
others.  This we can support through combinations of the layered
interfaces.

Charlie