[CBRD-21480] Do not allow concurrent computation of oldest active MVCCID

[alexstanro]

http://jira.cubrid.org/browse/CBRD-21480
When create index/upgrade/redistribute we obtain the latest oldest_active_mvccid. In order to do this, we are waiting for any concurrent thread that computes oldest_active_mvccid. In case of upgrade/redistribute we do not allow to advance with oldest_active_mvccid up to the end of transaction.
This is done with vacuum_Global_oldest_active_blockers_counter, having the following values:

• COMPUTE_OLDEST_MVCCID_STARTED - the current thread started update oldest_active_mvccid, the concurrent thread can't update oldest_active_mvccid (the concurrent thread uses the cached value or is blocked)
• greater than 0 - there is no thread computing oldest_active_mvccid, the threads can't update oldest
• 0 - one thread can update the oldest

[cubridci]

Build comment file:

File: src/query/vacuum.c (2 incorrect indentation(s) found)

  2567 	  ATOMIC_STORE_64 (&vacuum_Global_oldest_active_mvccid, local_oldest_active_mvccid);
  2568 	}

---

Please run 'indent -l120 -lc120' to correct the indentation

---