Now IR supports only reading full array column witouth working with indexes

Explain should show the same version of the selection clause as we
use to build SQL.

Now we support a flashback traversal for the subtree iterator. As
a result we can traverse the original selections in a way they looked
before any transformations were applied. This solves the problem
of too verbose DNFs that can easily reach tarantool's parser stack
limit.

Split different traversal iterators to separate modules.

As we use an append-only arena for the plan, we can easily implement
an UNDO log for each transformation we make. It can be helpful when
we would like to undo any of the transformations.

The main goal of it now - to rewind transformed selection to the
original state, as it can produce a logically equivalent, but too
long SQL string that breaks tarantool parser on the storages.

We have already fixed the problems with the bucket_id system column
(fc8b6d93), but forgot to take care
of the join condition references. Now this problem is fixed.

Now sbroad has distributed SQL functions in the core crate. At the
momnet there is only a single implementation of the `bucket_id()`
function.

Also add local SQL `bucket_id` function to router ans storage roles.

Though we can lazily initialize statistics spaces, there is still a
problem with read-only replicas. If the workload does not write to
the storage (only reads with SELECTs), then callbre is used and no
queries will be dispatched to the storage master. As a result, no
statistics spaces will be initialized on the storage failover group.

To deal with it, we add a new API method to initialize statistics
on the instance startup.

The query statistics samples the execution time of the opentelemetry's
spans among 1% of all the running queries. Collected data is stored
in the temporary spaces on each instance independently and can be
analyzed by user (with the local SQL) or exported somewhere.
The statistics spaces would be created by the sbroad itself if they
do not exist yet.