Dishonest schema reload timeout

There is a problem with this timeout:

local function call_reload_schema(replicasets)
    for _ = 1,replicasets_num do
        if channel:get(const.RELOAD_SCHEMA_TIMEOUT) == nil then
            for _, f in ipairs(fibers) do
                if fiber:status() ~= 'dead' then
                    f:cancel()
                end
            end
            return nil, ReloadSchemaError:new("Reloading schema timed out")
        end
    end
...
end

replicasets_num * const.RELOAD_SCHEMA_TIMEOUT may require much more time than const.RELOAD_SCHEMA_TIMEOUT

Originally posted by @oleg-jukovec in https://github.com/tarantool/crud/pull/319#discussion_r951224342

Start at call_reload_schema(replicasets) and inspect how replicasets_num and const.RELOAD_SCHEMA_TIMEOUT determine the reload loop's timeout. Trace the schema-reload fibers and look for existing tests or validation around reload failures. Done means the timeout behavior matches the intended overall operation and the timeout path remains covered.