add GSSAPI NAMETYPE socket options

Problem: zmq_gssapi.7 was not mentioned in doc/Makefile.am

Solution: add man page to MAN7 in doc/Makefile.am

Problem: new GSSAPI socket options are not documented.

Solution: add PRINCIPAL NAMES section to zmq_gssapi.7

Problem: principals are looked up unconditionally
with the GSS_C_NT_HOSTBASED_SERVICE name type.

Solution: Add two new socket options to set the name type
for ZMQ_GSSAPI_PRINCIPAL and ZMQ_GSSAPI_SERVICE_PRINCIPAL:

ZMQ_GSSAPI_PRINCIPAL_NAMETYPE
ZMQ_GSSAPI_SERVICE_PRINCIPAL_NAMETYPE

They take an integer argument which must be one of
ZMQ_GSSAPI_NT_HOSTBASED (0) - default
ZMQ_GSSAPI_NT_USER_NAME (1)
ZMQ_GSSAPI_NT_KRB5_PRINCIPAL (2)

These correspond to GSSAPI name types of:
GSS_C_NT_HOSTBASED_SERVICE
GSS_C_NT_USER_NAME
GSS_KRB5_NT_PRINCIPAL_NAME

Fixes #2542

RELICENSE: Hendrik Beskow grant 

fix misc. bugs in GSSAPI support

Problem: one call to gss_import_name() includes the terminating
NULL in a gss_buffer_desc.length, and one doesn't.

According to the examples at:
http://docs.oracle.com/cd/E19253-01/816-4863/overview-22/index.html
the NULL should be included in the length.

Solution:  Fix one case to include the terminating NULL in the length.

Problem: if client sets ZMQ_GSSAPI_PRINCIPAL to a name
for which credentials cannot be obtained, authentication
proceeds with default credentials.

Solution: Before initializing the security context, check
whether there was a failed attempt to acquire credentials
for a specific principal and bail out if so.

Fixes #2531

Problem: if client sets the ZMQ_GSSAPI_PRINCIPAL to a valid
principal, authentication fails.

When an application sets ZMQ_GSSAPI_PRINCIPAL, whether as a
client or a server, libzmq internally calls gss_acquire_cred()
with cred_usage=GSS_C_ACCEPT.  This cred_usage setting is for
acceptors (servers) only, thus it doesn't work for initiators
(clients).

Solution: Change the cred_usage parameter to GSS_C_BOTH to allow
initiators to set ZMQ_GSSAPI_PRINCIPAL.

Problem: the ZMQ_GSSAPI_PRINCIPAL socket option is described
as mandatory in the zmq_gssapi(7) manual page.  In fact it
is optional.

Solution: Describe ZMQ_GSSAPI_PRINCIPAL as optional.
If unspecified, default credentials are used.

Problem: configure.ac is not setting HAVE_LIBGSSAPI_KRB5
in src/platform.hpp when --with-libgssapi_krb5 is specified

Commit 09e868b7
switched the libgssapi_krb5 check from AC_CHECK_LIB
to AC_SEARCH_LIBS, but neglected to add an AC_DEFINE
for HAVE_LIBGSSAPI_KRB5, thus the GSSAPI code is
never compiled.

Solution: Add missing AC_DEFINE of HAVE_LIBGSSAPI_KRB5.

Problem: Windows build broken by #if ZMQ_USE_POLL

Solution: use #if defined ZMQ_USE_POLL

* change macOS < 10.12 clock to SYSTEM_CLOCK, fixes #2537

* remove clock_id option from alt_clock_gettime since we always want a monotonic clock.

* update header definition for alt_clock_gettime

* pass clock definition down to host_get_clock_service for macOS < 10.12

* change to monotonic clocks

Add CMake Build Output Options

Add two new options to CMakeLists.txt:

`BUILD_SHARED` - Whether or not to build the shared object (Default: ON)
`BUILD_STATIC` - Whether or not to build the static archive (Default: ON)

* Problem: Minor inconsistency in macro syntax. Solution: Code fixed.

Problem: FD set copying for Windows still not optimal in some places.

Problem: FD set copying on Windows still not optimal on some places. Solution: Improved memcpy() of FD sets for Windows builds.

Problem: Stack overflow in Windows VS 2012 builds for simple ZeroMQ u…

Problem: Stack overflow in Windows VS 2012 builds for simple ZeroMQ usage. Solution: Added notice in INSTALL file to mandatory use at least 2 MB stack size in VS 2012 and recommendation to use at least 2 MB in all other Windows builds.

RELICENSE: Pawel Kurdybacha grant

Fix compilation on windows using msys2

Problem: GCC 6 build fails due to misleading indentation and visually ambiguous if/else block ordering.

Solution: fixed indentation and if/else block.

RELICENSE: William Swanson

Critical BUG introduced with #2518 fixed.

Problem: Bad style in tcp.hpp/cpp

Solution: Add spaces between function
name and parentheses.

Another macro correction.

Consistency of Windows TCP tuning assertions

Treat ECONNRESET, ENETRESET and EINTR as non-fatal
failures in TCP Tuning on Windows as they are in the
Unixes.  Also adds ENETRESET to the Unix list that was missing.

Case found not covered in latest zmq::proxy() code.