Eliminate the ability to have multiple threads working on building the same message -- performance penalty is too large, and applies even to single-threaded users.

Revamp concurrency model, part 1:  EventLoop no longer allows cross-thread event queuing, simplifying many things.  Capability clients are no longer thread-safe, so they don't have to be so const.  In the future, explicit ways to communicate between threads will be re-added, but threads will be treated more like separate vats that just happen to have a particularly fat pipe.  Upcoming:  Remove mutexes.

Spin off libraries kj-async and capnp-rpc, so that people who don't want them can avoid their large footprint.

Refactor capability code using fork.  Still too much refcounting, though.  Maybe this calls for a different design for pipelining...

Interface code generation.  Also update method grammar to allow multiple returns, and allow specifying request/response structs instead of parameter lists.

Mark every destructor "noexcept(false)" unless it is very clearly noexcept.  The change in C++11 making destructors default-noexcept was a mistake.

totalSizeInWords() method on generated struct readers/builders returns total size of the struct, not counting far pointer overhead.  Useful for allocating space for a flat copy.  Also a couple misc tweaks.

Get rid of redundant ErrorReporter that goes in ReaderOptions.  Registering an ExceptionCallback is a better approach.

Replace most CAPNPROTO_ASSERTs with the new logging/error macros.  Also add a SYSCALL macro and use it.  Also add commentary.