InnoDB never uses fil_system_t::extend_to_recv_size() during recovery

While applying the redo logs, InnoDB has fil_system.extend_to_recv_size() to extend
the tablespace. fil_space_t::prepare_acquired() extends the tablespace while
opening the tablespace for write. In case of single scan of redo log records, InnoDB
has space->is_closing() return true and so it doesn't use extend_to_recv_size().
In case of multiple scan, InnoDB does fil_space_get() to know whether the
tablespace exists or not.