Use strcmp for string comparison.

On my laptop, with gcc -O, these comparisons didn't work as intended,
and as result, the FTS probe messages were not processed, and
gp_segment_configuration always claimed the mirrors to be down, even though
they were running OK.

src/backend/fts/fts.c

@@ -686,14 +686,14 @@ FtsWalRepSetupMessageContext(fts_context *context)
 	for (i = 0; i < context->num_primary_segments; i++)
 	{
 		probe_response_per_segment *response = &context->responses[i];
-		if (response->message == FTS_MSG_PROBE)
+		if (strcmp(response->message, FTS_MSG_PROBE) == 0)
 		{
 			response->message = NULL;
 			response->isScheduled = true;
 		}
 		else
 		{
-			Assert(response->message == FTS_MSG_SYNCREP_OFF);
+			Assert(strcmp(response->message, FTS_MSG_SYNCREP_OFF) == 0);
 			response->isScheduled = false;
 			response->result.isPrimaryAlive = false;
 			response->result.isInSync = false;

src/backend/fts/ftsprobe.c

@@ -213,10 +213,10 @@ ftsReceive(FtsConnectionInfo *ftsInfo)
 	 * response. In that situation, we should force another probe to update
 	 * the gp_segment_configuration to avoid waiting the fts probe interval.
 	 */
-	if (ftsInfo->message == FTS_MSG_PROBE)
+	if (strcmp(ftsInfo->message, FTS_MSG_PROBE) == 0)
 		probeRecordResponse(ftsInfo, lastResult);
 	/* Primary must have syncrep disabled in response to SYNCREP_OFF message. */
-	AssertImply(ftsInfo->message == FTS_MSG_SYNCREP_OFF,
+	AssertImply(strcmp(ftsInfo->message, FTS_MSG_SYNCREP_OFF) == 0,
 				PQgetvalue(lastResult, 0, Anum_fts_message_response_is_syncrep_enabled) != NULL &&
 				*(PQgetvalue(lastResult, 0, Anum_fts_message_response_is_syncrep_enabled)) == false);