Without this change, a segmentation fault is likely to occur when using the proposed snippet of code, as `*address[size]` is equivalent to `*(address[size])`, not `(*address)[size]` as clearly intended.

Problem: drop tests on CLIENT/SERVER are misleading

- both are thread safe
- they do not accept ZMQ_SNDMORE nor ZMQ_RCVMORE

These sockets deal only with single part data (at present), so the
drop tests are misleading / inappropriate.

Solution: remove them.

Various cleanups for CLIENT-SERVER sockets

REQ-REP barely work, and DEALER-ROUTER are misnamed. As we move to
formalizing CLIENT-SERVER, we can deprecate request-reply.

Solution: move this section to the end of the man page, and mark as
"will be deprecated".

Solution: add man pages/sections for:

 * ZMQ_SERVER
 * ZMQ_CLIENT
 * ZMQ_THREADSAFE
 * zmq_msg_routing_id
 * zmq_msg_set_routing_id

Solution: check/reject a zero routing ID with EINVAL.

These tests connected CLIENT and SERVER to DEALER... this isn't
allowed. I changed to CLIENT-to-SERVER in both cases. The result
was aborts in client.cpp and server.cpp which cannot handle
invalid multipart data.

I removed the asserts in each of these in xsend.

Solution: fix the test cases and remove the (unwanted?) asserts
in client.cpp:xsend and server.cpp:xsend.

Solution: fix to use CLIENT-to-SERVER and increase number of messages
sent to 100K per thread, to better stress thread safety.

This is due to the mangled include of platform.h, which was to make
CMake happy.

Solution: in CMakeLists.txt, define USING_CMAKE and then look for
platform.h in current directory if that is defined, else look in
../src/ as one would expect.

It should never be zero.

Solution: if rid cycles back to zero, bump to 1.

This has no sense IMO; people who need this can still use DEALER-DEALER.

Solution: let CLIENT talk only to SERVER.

We do not use 'get' for getters.

Solution: rename to zmq_msg_routing_id () for public API.

avoid dereferencing uint32_t on unaligned address

Solution: fix typos

Problem: Tests in Travis are failing randomly

Tests were failing, because some deque calls were causing undefined
behavior: calling front() or pop_front() on an empty deque. Such
calls are now safeguarded.

Fixed missing subscriptions on XPUB with manual subscriptions

The patch fixes the issue #1568.

Problem: lingering subscriptions on XPUB sockets (#1566)

The patch fixes lingering subscriptions that occur upon disconnection
on XPUB sockets with option XPUB_MANUAL when used in a XPUB-XSUB
proxies.

Solution: do not call poller->add_fd() with -1. Fixes #1558

Revert "Solution: check return code of make_fdpair() with zmq_assert"

Make DragonFly a first class citizen in terms of compilation

Solution: check return code of make_fdpair() with zmq_assert

Cleaned up test case

Update zmq_inproc manpage for bind/connect order

The order of zmq_bind() and zmq_connect() is no longer important
for the inproc transport since libzmq 4.0.  This updates this
in the zmq_inproc manpage.

Problem: Identity frame from router has no metadata

Problem: Private/internal macros defined in public header.