* Remove pcall_wrapper()
* Use reflection to implement err:__index() and err:unpack()

Breaking changes:

 * pcall() returns 'struct error' cdata instead of string for box.error()
 * box.error.last() returns 'struct error' instead of Lua table

Refactor engine.cc to make a snapshot
based on existing, not future, WAL state.

Implement wal_checkpoint() which delivers WAL
vclock to tx thread and reopens the WAL.

* Remove recovery->apply_row and recovery_apply_row
* Remove struct recovery from sophia_engine
* Prepare to remove struct recovery from memtx_engine
* Prepare to remove struct recovery from applier

Delete the previous checkpoint in sophia only if we have a new one.

Actually fix the bug and add a test case.

Actually fix the bug and add a test case.

Conflicts:
	src/box/memtx_engine.h
	src/box/sophia_engine.h

Conflicts:
	src/box/lua/schema.lua

This makes the engines less dependent on the global recovery
object (eventually this dependency will be removed altogether).

* introduce MemtxEngine::m_snap_dir
* begin using it during recovery
* in MemtxEngine::join() create a separate xdir object,
  to avoid races (the method is invoked from 'relay' thread)
* remove recovery_setup_panic() and pass panic_on_wal_error
  to recovery_new().
* remember the last checkpoint in memtx engine state and
  use it for replication join calls (rather than xdir_scan()
  to find the latest snapshot).
* remove recovery_has_data()
* move recovery_last_checkpoint() to memtx_engine.cc
* remove recovery::snap_dir
* use the vclock from memtx engine in join
* now a different error message is thrown
for the case when snapshot file is missing

Added override keyword for proper methods of index and engine derivative classes

closes gh-1428

Use a global variable with server id.

* do not use snap_dir to load bootstrap.snap, just any dir object
  will do
* move snap_io_rate_limit to memtx_engine
* remove box_set_panic_on_wal_error(), the dynamic
  setting is used for relays, there is no point
  to update the global recovery instance after box.cfg{}

Fixed double call of tuple_init_field_map in update and upsert operations, fixed comments of tuple_alloc

If `default_cfg` was nil, then system will errored.

This is necessary to ensure the request is alive until
request_encode(), which will be moved to WAL thread in the future.

* Don't mislead users with box.cfg{} options which shouldn't be
  set under tarantoolctl.
* Mention that all data directories must have absolute paths and
  remove work_dir from example.lua (see #1406).
* Specify correct default values.

+ closes gh-1405 (right path for unix socket)
+ wrap box.cfg and loadfile with pcall
+ some variable/function scope cleanups (global -> local)
+ closes gh-1409 (check for lua syntax before running/restarting/eval server)

Ensure that every box operation can take tuple as key/tupl/operation in
both engines