A pre-requisite patch for data dictionary: sotre index
ordinal number in the key_def object.
Remove struct space from index constructor.

1) Lua Tarantool Server, that runs plain lua code and admin server commands (separated ';\n')
2) Python Tarantool Server, that runs old plain tests.
Also created Two types of tests, that inherit FuncTest (Cause they're FuncTest):
1) LuaTest -> Lua Tarantool Server
2) PythonTest -> Python Tarantool Server
In the future we plans to add luabench as one of Test types
Rename all cores to appropriate names of server (now it's 'lua tarantool' and 'python tarantool')
Prepare to remove lua prefixes (if you need to test wrong or nonimplemmented command you may run 'admin('cmd', simple=True)')
If you want to implement new server command - add needed regexp to admin_connection.py.
if it's fixed length command, then add it to rg1, else rg2

Incompatible change,
https://blueprints.launchpad.net/tarantool/+spec/cardinality-arity

cardinality is how many unique elements a set contains. wrong term
for the max number of fields in a tuple

https://bugs.launchpad.net/tarantool/+bug/1200228
Remove the regression I introduced when trying to be to smart when
comparing integer numbers.

Numeric comparison is not memcmp compatible on x86, and I don't
know what I was smoking when I wrote a test which proved it wrong.

1) Remove debug output
2) Add --gdb description
3) Return [self.binary]

Add forgotten tarantool error.

1) Moving ./test-run.py --gdb from xterm (or another standart X terminal emulator) to detached Screen session
2) Moving from unnecessary dependency "pexpect" to builtin "subprocess" module.

Move struct key_def into an own header.
Rename key_def methods to follow the coding style.
Rename field_data_type to field_type.
Instead of panicking on key_def initialization
error, throw an exception.

The allocator consists of a common slab cache, region and slab
allocators.
The patch provides unit tests.

It also modifies rlist_shift to be more in line with the
rest of rlist functions.

Derive some tests for multipart HASH indexes
from the original implementation by A. Demakov
before dropping the branch.

Don't drop entire vardir on test suite start, it may have
symlinks, configuration files, etc. - useful for debugging.

1) Temporary result files weren't deleted, if test's skipcond failed.
2) Vardir is removed when tests started.
3) Sort tests and test-suites by name
4) Added cleanup to box/lua.test

* memcached/expirations.test
* memcached/flush-all.test
* replication/memcached.test