[PATCH 2/7] winedbg: Simplify and fix register read/write handlers.

[Rémi Bernon]

It looks like that while rebasing/cleaning up the patches before sending 
them I messed with the precondition checks, and thread->process->be_cpu 
is accessed before the !thread if verified...

The thread should never be NULL in any case, and gdb frontend does not 
seem to handle error replies very well, so maybe assert is in order 
rather than checking and reply with error. Not sure what's preferred here?
-- 
Rémi Bernon <rbernon at codeweavers.com>