[PATCH 2/2 weston] toytoolkit: Don't draw shadows for maximized windows.

[Bill Spitzak]

This looks really messy and still makes the assumption that the only 
reason for "edges" to be removed is because of "maximize". It is also 
completely different than how "fullscreen" is done, even though that 
should be identical except the "titlebar" is also removed for fullscreen.

Based on the latest changes to the shell api, I feel the following must 
be added:

1. The "content" region. This is the part of the window that does not 
include the "edges". However it *does* include the titlebar. Imagine the 
"part of the window you see when it is maximized". This is different and 
smaller than the input region and should be specified by clients using a 
new api very similar to how the input and opaque regions are specified. 
Shells need this information to property implement snap-to-edge (where 
the edge is of another window, a panel, or an output), and to implement 
maximizes other than the full-screen maximize (ie vertically-only). It 
also means "maximize" is not a special state, instead the shell just 
resizes the surface so that the contents fill the output.

2. When the shell tells a client to configure a window, it should 
communicate both a size for the surface image, and the bounding box of 
the content region in this new size. This bounding box will always be 
the same distance or closer to the edges of the surface (ie it is only 
used to clip edges, not to make them thicker). This allows clients to 
know that edges are clipped so they do not place any important controls 
there, and they can adjust their graphics to take into account the 
clipping (for instance not drawing rounded-corner shading for a 
clipped-off rounded corner). It will also save some memory and rendering 
time by clipping these off the allocated image buffer.

Ideally this new box should be added to the configure event and we 
should just require clients to obey it. If this is not allowed due to 
the api being frozen, it could be a different event, but it suffers from 
the ugly fact that the size sent with configure is not the actual size 
the surface should be (because that would make a client that does not 
clip the edges make the window too small when maximized).

3. Fullscreen: somewhat unrelated, but the shell should be able to 
directly tell clients to "remove all decorations". Currently this is 
tied into the fullscreen request from a client. This is necessary so 
shells can force a client to fullscreen. It is also needed for embedding 
one wayland client in another. It may also be useful for displaying 
wayland clients on legacy window systems that make it very difficult to 
remove their own decorations.