Remove log_io_iter from snapshot IO. Do not use coroutines
and extra buffers. This opens the way to more agressive
buffering of snapshot IO, as compared to no buffering of WAL IO.

Cleanup work with eof/row markers. Remove unnecessary members
of log_io_class.

Fix a regression introduced by the yesterday's cleanup:

recover_wal() can return 0, -1, or 1:
0 - ok, eof read
-1 - error reading wal
1 - can't read eof marker from wal.

recover_remaining_wals() looks at this, and when doing final recovery
deletes wal which don't have eof marker (return value from recover_wal()
== 1).

Yesterday I mistakenly merged 0 and 1 return values into one.
Revert this part of the yesterday's patch.

On the same token, log_io_cursor_next() can't throw an exception,
ever. Take this into account and simplify code.

recover_remaining_wals() continues to stay a mess.

- remove unused parameters
- add SystemError exception (derived form net-split tree)
- improve error printouts.

Conflicts:
	include/tarantool_lua.h
	src/tarantool_lua.m

Adjust naming, an error message, don't use a global function  where
a static one would make do.

* using new delete version with flags
* splice offset type fixed
* added tnt_tuple_reply tuple validation + test case

The nullify potential unsafe (in security mind) functions was moved from box.lua
to separate functions which calls after init.lua. It allows call require,
os.execute, and another unsafe function in init.lua.

Add Request object with method execute. Allocate
Requests on fiber->gc_pool. Clear txn from request data.
Make txn commit/rollback algorithm independent on the actual
request or request data.
Update valgrind suppressions with new call paths.