Some code in libldap which we don't need requires libresolv,
so we should relax linking rules to let linker drop it.

NB: this commit needs the fixes in tarantool-sys,
so we have to update the git submodule.

Previously we extracted attributes from DDL plan nodes and left
the nodes in arena in the inconsistent state. This approach is
error prone, so now we extract the whole node from the plan arena
before using its attributes.

Previous version of the SQL module hierarchy was overcomplicated.
Current commit simplifies it, so SQL has only two submodules:
router and storage.

Previously we created a very strange error sandwitch when failed
on the storage or router runtimes:
- wrapped traft error into a sbroad one
- wrapped the result again into traft error

Now we wrapp sbroad errors into traft or return traft errors as is.

Now it is possible to drop sharded tables with pico.sql() Lua
function.

Adds support for creation of the sharded memtx spaces via pico.sql().

The reason of this refactoring is that we need Picodata specific
API rather then the API provided by tarantool module. For example,
table creation needs access to the clusterwide API for sharded and
global spaces.

Georgy Moshkin authored 1 year ago and Georgy Moshkin committed 1 year ago

Closes #321