Solution: automate it on Travis

Problem: race condition breaks automated OBS release

Solution: instead of using a temporary release branch that gets
deleted after an arbitrary timer, which never matches the time OBS
actually takes to do a full git clone, keep the branch permanently
in place and simply move it forward to the latest tag on each
release.

Problems: test failures on PPC and GNU/hurd

Solution: mark it as XFAIL like the other tests that use ipc://*

Solution: prefer CXX11 atomics if they are available to compiler
intrinsics.
test_hwm_pubsub fails 50% of the times on PPC64 and RISC-V with an
apparent memory corruption of messages sent by the application thread
to the I/O thread when using compiler intrinsics.
Switching to CXX11 atomics makes the test reliable again. The
standard API should be preferred anyway, if available.

Solution: import macro from autoconf-archive that works better
than AC_CHECK_FUNCS

Solution: detect cacheline size for aligment purposes at build time
instead of hard-coding it, so that PPC and S390 can align to a value
greater than the 64 bytes default.
Uses libc getconf program, and falls back to the previous value of 64
if not found.

Problem: ZMQ_SOCKET_LIMIT and ZMQ_THREAD_PRIORITY have the same value

Solution: remove documents and tests for ZMQ_THREAD_PRIORITY getter. It
never worked and can never work as it has the same value as a get-only
option ZMQ_SOCKET_LIMIT. It cannot be changed without breaking ABI.
Note that the setter works fine as ZMQ_SOCKET_LIMIT is get-only.

Problem: GCC warnings about string overflows in tests

Solution: mostly false positives, but fix them

Problem: gssapi pkg-config check in configure.ac does not work

Solution: correctly enable the definition in platform.hpp so that the
gssapi support is actually built in if requested and available.

Problem: NEWS for 4.3.1 does not mention CVE number

Solution: add it now that it's been assigned

Problem: test_rebind_ipc still fails

Solution: create manually the random IPC file to reuse

Problems: test failing on SPARC64 and hard-coded socket binds

Solution: use wildcards or random directories to avoid races when
multiple users are running the same test on the same machine

Solution: use memcpy instead of doing pointer arithmetics with casting
and dereferencing to fix the error on sparc64

Problem: versions are out of date

Solution: bump ABI and revisions and changelog now that v4.3.1 is out

Problem: NEWS out of date with recent changes

Solution: update it

Problem: small formatting issue highlighted by clang

Solution: fix it with make clang-format-diff

V2 decoder overflow RCE fix

Solution: Cast the signed expression (which is always positive) to unsigned
Signed-off-by: Guido Vranken <guidovranken@gmail.com>

Solution: added relicense agreement
Signed-off-by: Guido Vranken <guidovranken@gmail.com>

Problem: pointer overflow in zmq::v2_decoder_t::size_ready leading to remote code execution (issue #3351)

Solution: refactor bounds check arithmetic such that no overflow shall occur
Signed-off-by: Guido Vranken <guidovranken@gmail.com>

Problem: unused import and inconsistent use of const

Solution: remove unused import and fix const placement.

Problem: race condition in PUSH might break atomicity

Solution: if a multi-part message cannot be delivered by a PUSH socket
because the pipe is broken mid-way, drop the rest of the frames so that
the atomicity property is not broken.

Problem: tests without test framework

Solution: migrate to unity

Problem: Android helper does not redirect pkg-config

Solution: set the appropriate options, environment variables and paths
so that the host pkg-config files are ignored, and the target ones are
used instead (if any)