Replace list() with set() validation in analyzedb

After testing analyzedb on a huge database with 170k tables we
have found a bottleneck while printing candidate list to analyze.
It took about 45 minutes to print all tables. The bottleneck was
in O(n^2) complexity when we validated candidates in a loop with
a list() instead of set(). The same O(n^2) validation is made while
running analyze commands on executor pool.
This commit change candidate type from list() to set() to reduce
complexity from O(n^2) to O(n).

gpMgmt/bin/analyzedb

@@ -24,6 +24,7 @@ from datetime import datetime
 from contextlib import closing
 import pipes  # for shell-quoting, pipes.quote()
 import fcntl
+import itertools
 
 try:
     import pg
@@ -616,12 +617,12 @@ class AnalyzeDb(Operation):
         3. skip views and external tables
         """
         qresult = run_sql(self.conn, GET_MID_LEVEL_PARTITIONS_SQL)
-        mid_level_partitions = []
+        mid_level_partitions = set()
         for schema_tbl in qresult:
             tup = (schema_tbl[0], schema_tbl[1])
-            mid_level_partitions.append(tup)
+            mid_level_partitions.add(tup)
 
-        ret = []
+        ret = set()
         for can in candidates:
             schema = can[0]
             table = can[1]
@@ -632,16 +633,16 @@ class AnalyzeDb(Operation):
             if can in mid_level_partitions:
                 logger.warning("Skipping mid-level partition %s.%s" % (schema, table))
             else:
-                ret.append(can)
+                ret.add(can)
 
         if self.config_file is not None or self.single_table is not None:
-            valid_tables = []
+            valid_tables = set()
             if len(ret) > 0:
                 oid_str = get_oid_str(ret)
                 qresult = run_sql(self.conn, GET_VALID_DATA_TABLES_SQL % oid_str)
                 for schema_tbl in qresult:
                     tup = (schema_tbl[0], schema_tbl[1])
-                    valid_tables.append(tup)
+                    valid_tables.add(tup)
             return valid_tables
 
         return ret
@@ -856,7 +857,7 @@ class AnalyzeDb(Operation):
         2. The leaf partitions (if range partitioned, especially by date) will be ordered in descending
            order of the partition key, so that newer partitions can be analyzed first.
         """
-        candidate_regclass_str = get_oid_str(candidates + list(root_partition_col_dict.keys()))
+        candidate_regclass_str = get_oid_str(itertools.chain(candidates, root_partition_col_dict.keys()))
         qresult = run_sql(self.conn, ORDER_CANDIDATES_BY_OID_SQL % candidate_regclass_str)
         ordered_candidates = []
         for schema_tbl in qresult: