Fix MVCC bug with prepared xact with subxacts on standby

We did not recover the subtransaction IDs of prepared transactions
when starting a hot standby from a shutdown checkpoint. As a result,
such subtransactions were considered as aborted, rather than
in-progress. That would lead to hint bits being set incorrectly, and
the subtransactions suddenly becoming visible to old snapshots when
the prepared transaction was committed.

To fix, update pg_subtrans with prepared transactions's subxids when
starting hot standby from a shutdown checkpoint. The snapshots taken
from that state need to be marked as "suboverflowed", so that we also
check the pg_subtrans.

Backport to all supported versions.

Discussion: https://www.postgresql.org/message-id/6b852e98-2d49-4ca1-9e95-db419a2696e0@iki.fi

src/include/storage/standby.h

@@ -75,11 +75,19 @@ extern void StandbyReleaseOldLocks(TransactionId oldxid);
  * almost immediately see the data we need to begin executing queries.
  */
 
+typedef enum
+{
+	SUBXIDS_IN_ARRAY,			/* xids array includes all running subxids */
+	SUBXIDS_MISSING,			/* snapshot overflowed, subxids are missing */
+	SUBXIDS_IN_SUBTRANS,		/* subxids are not included in 'xids', but
+								 * pg_subtrans is fully up-to-date */
+} subxids_array_status;
+
 typedef struct RunningTransactionsData
 {
 	int			xcnt;			/* # of xact ids in xids[] */
 	int			subxcnt;		/* # of subxact ids in xids[] */
-	bool		subxid_overflow;	/* snapshot overflowed, subxids missing */
+	subxids_array_status subxid_status;
 	TransactionId nextXid;		/* xid from TransamVariables->nextXid */
 	TransactionId oldestRunningXid; /* *not* oldestXmin */
 	TransactionId oldestDatabaseRunningXid; /* same as above, but within the