Before the patch, big integer constants were silently
converted to floating point values. Fix that by issuing
error message if value doesn't fit in int64_t range.

Also, refactor surrounding code as per Tarantool's code style.

Closes #2347

Now every sqlite struct Index is created with tnt
struct index_def inside. This allows us to use tnt
index_def in work with sqlite indexes in the same
manner as with tnt index and is a step to remove
sqlite Index with tnt index.
Fields coll_array, coll_id_array, aiColumn,
sort_order, aiRowLogEst and zName are removed from
Index. All usages of this fields changed to usage of
corresponding index_def or index_def->opts fields.
index_is_unique(), sql_index_collation() and
index_column_count() are removed with calls of
index_def corresponding fields.
Also there is small change in behavior: before the
patch a statement like
CREATE TABLE t1(a,b, PRIMARY KEY(a,b), UNIQUE(a,b))
created only one constraint index (for primary key)
and no index for UNIQUE constraint (since it is upon
the same columns), neither it is named or non-named
constraint.
After the patch index will be always created for named
constraints. It is a temporary solution. In future
it's preferable not to create an index, but to make
some record in _constraints space that this named
unique constraint implemented with the same index as
primary key constraint.

Closes: #3369, #3498

Binary search in whereKeyStats function that uses routine
sqlite3VdbeRecordCompareMsgpack returned invalid result,
because it made wrong assumption about the number of
coinciding index columns based on _sql_stat4.
But all collected index statistics were wrong.

The analyzeOneTable routine used to call sqlite3GetTempRange
and sqlite3ReleaseTempRange functions inside of cycle intensively
changes Parser->nMem value. Thanks this pattern routine
sqlite3ReleaseTempRange could push nMem variable onto aTempReg
queue. As Parser->nMem could be bigger next time, this
variable may be used twice:

          [74, 'Column', 4, 1, 11, '', '00', 'r[11]=']
          [75, 'Column', 4, 2, 12, '', '00', 'r[12]=']
|->       [76, 'Column', 4, 3, 13, '', '00', 'r[13]=']
          [77, 'Column', 4, 4, 14, '', '00', 'r[14]=']
!rewrite  [78, 'Column', 4, 0, 13, '', '00', 'r[13]=ID']
          [79, 'MakeRecord', 13, 1, 6, '', '00',
           'r[6]=mkrec(r[13])']

Thus all assumptions about data below should be invalid and
some artifacts could exhibit.

Example:
\set language sql
CREATE TABLE t1(id INTEGER PRIMARY KEY AUTOINCREMENT, a,
		b, c, f, d, e);
CREATE INDEX i1 ON t1(a, b, c, f, d);
\set language lua
for i = 0, 100 do
    box.sql.execute(string.format("INSERT INTO t1 VALUES(null,
		    'x', 'y', 'z', 'a', %s, %s);", i,
		    math.floor(i / 2)))
end
\set language sql
SELECT * FROM "_sql_stat4";
---
['T1', 'I1', '101 101 1 1 1', '0 0 44 44 44', '0 0 44 44 44',
 ['x', 'y', 'z', 'a', 44]]
---

Expected:
---
['T1', 'I1', '101 101 101 101 1', '0 0 0 0 44', '0 0 0 0 44',
 ['x', 'y', 'z', 'a', 44]]
---

Resolves #2847.

If is_first_insert flag is set and vy_stmt_type(rv->tuple) equals
IPROTO_DELETE, we free rv->tuple, but then we dereference it via
an on-stack variable to check if we need to turn a REPLACE into an
INSERT or vice versa. Fix this.

This is not necessary, as we can use fiber()->gc, as we usually do.

Move key_def creation to compare_write_iterator_results as it is the
same for all test cases. Performance is not an issue here, obviously, so
we can close our eyes to the fact that now we create a new key def for
each test cases.

The new function inserts a new item into the list at the specified
postion.

If a fiber pool reuses already canceled fiber then the fiber reports an
error for any next request. Now canceled fiber returns and fiber pool
creates a new one.

Fixes #3527

Disallowed division by zero in SQL to follow ANSI standard.
This also affects some unobvious scenarios when divider indirectly
casts to zero like
SELECT 'word1' / 'word2';

Closes #2135.

New commit in third_party/libyaml downgrades
required cmake version.

If specified PK column modifier refers nonexistent column,
lookup by name keeps NULL column pointer unchanged that should
be accounted in assert.

Close #3522.

Closes #2946.

@TarantoolBot document
Title: fixed module reload
There was a bug in tarantool documentation:
https://tarantool.io/en/doc/1.7/book/box/
box_schema/#lua-function.box.schema.func.reload
Now it is allowed to reload all functions in loadable
module via one method. Legacy method including finction
name is forbidden.
box.schema.func.reload("utils")       -- ok since now
box.schema.func.reload("utils.func1") -- forbidden since now

Global reload is still unsupported because it seems
to be useless.
box.schema.func.reload()              -- invalid!

Need to update tests as with fixup in upstrem
commit baf636a74b4b6d055d93e2d01366d6097eb82d90
Author: Tina Müller <cpan2@tinita.de>
Date:   Thu Jun 14 19:27:04 2018 +0200

The closing single quote needs to be indented...
if it's on its own line.

Closes #3275.

Closes #3275.

Found by @ImeevMA

 - Factor out local_recovery() from box_cfg_xc(). Make it setup
   replication and handle local recovery and hot standby cases.
 - Move replication setup in case of initial bootstrap from box_cfg_xc()
   to bootstrap() to make bootstrap() consistent with local_recovery().
 - Move initial snapshot creation from bootstrap() to bootsrap_master()
   and bootstrap_from_master().

Needed for #461

box_sync_replication() can now be called before recovery, right after
box_listen(). This is a step toward detecting if the instance fell too
much behind its peers in the cluster and so needs to be rebootstrapped.

Needed for #461

In order to find out if the current instance fell too much behind its
peers in the cluster and so needs to be re-bootstrapped we need to
connect it to remote peers before proceeding to local recovery. The
problem is box.cfg.replication may have an entry corresponding to the
instance itself so before connecting we have to start listening to
incoming connections. So this patch moves the call to box_listen()
before recoery is started unless the instance in hot standby mode.
It also folds box_bind() into box_listen() as it is no longer needed
as a separate function.

Needed for #461

In order to find out if the current instance fell too much behind its
peers in the cluster and so needs to be rebootstrapped, we need to know
its vclock before we start local recovery. To do that, let's scan the
most recent xlog. In future, we can optimize that by either storing end
vclock in xlog eof marker or by making a new xlog on server stop.

Needed for #461

Introduce vclock_is_set() helper and use it on xlog_meta::prev_vclock
instead.

Follow-up ac90b498 ("xlog: store prev vclock in xlog header").

This opcode was used when Vdbe had to store key_def in P4 for
OP_OpenRead/Write, so P4 was occupied and OP_LoadPtr was used to
store space pointer in another opcode. But now P4 is free for
OpenRead/Write and space pointer can be stored here.

Alongside, some useless key_def duplications are removed.

Made vdbe_emit_open_cursor calls consistent:
now it uses index id everywhere.
This required to change a way to detect that
VDBE has openned Read cursor to specified table
in vdbe_has_table_read to write result of insert
in temp table if required.

As sqlite3NestedParse became totaly useless, let's
remove unreacheble code and all mentioning.

Resolves #3496.

Now we manually generate AST structures to drop outdated
stats from _sql_stat1 and _sql_stat4 spaces instead of
starting sqlite3NestedParse. This function becomes totally
useless and could be removed.

Part of #3496.

Store replication group id in vylog and ignore spaces whose group_id
equals GROUP_LOCAL when relaying initial join rows.

Folow-up #3443

Follow-up #3443

Replica local spaces can't cause replication conflicts so we don't
really need to respect box.cfg.read_only for them (they are similar
to temporary spaces in this regard).

Follow-up #3443

Next checkpoint time is set by the formula:
 period = self.checkpoint_interval + offset,
where offset is defined as follow:
 offset = random % self.checkpoint_interval

So offset must be calculated again if at least the new
interval is less than the old one.

Closes #3370