* add proctitle.h;
 * add config-time detection of setproctitle() syscall;
 * fixing program_invocation_(short)?_name globals (glibc);
 * fixing progname (getprogname(), initialized with argv[0]);
 * this module now works on osx;
 * fix osx ps displaying garbage (argc-1 extra \0 necessary);
 * general refactoring;
 * fix linux proctitle truncation (init_set_proctitle was called with
   argv+1, this is wrong);

 * CAVEAT: 'progname: ' prefix no longer present, will fix later

Heading style suggestion:
Fix overlapping (e.g. http://tarantool.org/doc/getting_started.html#starting-another-tarantool-instance-and-connecting-remotely)
Proper margins

Codeblock style suggestion:
Add greyish background for inline code
Add background for bash codeblocks
Add overflow property for bash codeblocks:

Add horizontal scrolling for text in code-blocks

Fix multiple issues listed in #1033 (Database section):
Addressed style inconsistency
Fixed module descriptions
Added hotlinks for functions
Minor style additions

Fix formatting issues in Reference Manual

Fix Lua code highlighting

A rather blunt fix that offers a temporary workaround.
The reason behind original Lua lexer not highlighting Lua code is '!' used as
delimiter. Pygments fails to parse '!' as a Lua code entity and tells Sphinx to
leave the code unhighlighted. The fix adds custom Lua lexer by alias 'lua_tarantool',
with '!' added as a punctuation token (line 60 in LuaLexer.py). A proper fix would
probably be to extend the lexer with all Tarantool extensions over standard Lua modules.
Lua code block directive now should be used like this:

.. code-block:: lua_tarantool

	lua code

Fix remaining problems from #1033

Review styles for Terminal session and inline code

Improve formatting for C API

Improve formatting in Lua tutorial

This reverts commit 2d8b1416.

TODO: pass exact lsn as row.lsn to a slave during JOIN.
Ensure that slave properly accepts rows with variadic lsn.

tuple_upsert_execute() is designed to work with
region allocator, in other words it does not free
allocated memory during work.

Simulate region allocator behaviour in allocator stub.

In future this should be fixed using cords instead of
sophia thread.

Schedule background read operation only in cases when
cursor cache does not contain a key.

Fix memory leak issue for a highly loaded instances.

* upgrade isolation level to SSI (Serialized Snapshot Isolation)
* include read operation in transactional context (no cursor support)
* port hermitage tests (isolation certification)

The txn API must let the engine know when the first
statement happens in a multi-statement transaction.

For autocommit mode, statement transaction management
of read-only statements is just a thin layer around
fiber_gc().

Speed up txn construction and make txn object smaller:
* initialize triggers in txn on first use
* remove unnecessary members of txn
* replace rlist with slist (double linke list with single linke list)
  for statement list; implement a nicer single linked list for
  that purpose, and cover it with tests;

Streamline engine API:
* replace beginStatement with begin(),
  engines have no need to be alerted on each statement
* don't commit statement transaction inside the engine,
  do it once outside engine API
* TupleGuard -> TupleRef, implement TupleRefNil

Connection was declared successfully established prematurely; broke
tests on mac.

Conflicts:
	src/box/memtx_engine.cc
	src/lib/small