create signaler::recv_failable()

In real world usage, there have been reported signaler failures where the
eventfd read() or socket recv() system call in signaler::recv() fails,
despite having made a prior successful signaler::wait() call.

this patch creates a signaler::recv_failable() method that allows
unreadable eventfd / socket to return an error without asserting.

fix error with gcc 5.2

Solution: full is if number of msg is greater than hwm, not greater than hwm - 1

Solution: use existing check_hwm() method

Update cmake to support Visual Studio 2015

Fix typo on send documentation re: nullification

fixed zmq assertion in signaler.cpp under ubuntu

Fixed the bug which cause the application crashed when use TCP connec…
Fixes #1581 

Fixed the bug which cause the application crashed when use TCP connect to an invalid address(such as 192.168.1.255).

Fix windows cmake build

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