feat(jdbc): improve JDBC PostgreSQL queues queries performance (#1883)

close #1121

jdbc-h2/src/main/resources/migrations/h2/V18__index_logs.sql

@@ -0,0 +1,3 @@
+DROP INDEX logs_namespace;
+DROP INDEX logs_timestamp;
+CREATE INDEX logs_namespace_flow ON logs ("deleted", "timestamp", "level", "namespace", "flow_id");

jdbc-postgres/src/main/resources/migrations/postgres/V22__index_queues.sql

@@ -0,0 +1,24 @@
+-- Recreate the queues_type__* indexes by adding the offset column otherwise the index is not used as we order on offset.
+-- Also make them partial to lower the index size.
+DROP INDEX queues_type__consumer_flow_topology;
+DROP INDEX queues_type__consumer_indexer;
+DROP INDEX queues_type__consumer_executor;
+DROP INDEX queues_type__consumer_worker;
+DROP INDEX queues_type__consumer_scheduler;
+
+CREATE INDEX queues_type__consumer_flow_topology ON queues (type, consumer_flow_topology, "offset") WHERE consumer_flow_topology = false;
+CREATE INDEX queues_type__consumer_indexer ON queues (type, consumer_indexer, "offset") WHERE consumer_indexer = false;
+CREATE INDEX queues_type__consumer_executor ON queues (type, consumer_executor, "offset") WHERE consumer_executor = false;
+CREATE INDEX queues_type__consumer_worker ON queues (type, consumer_worker, "offset") WHERE consumer_worker = false;
+CREATE INDEX queues_type__consumer_scheduler ON queues (type, consumer_scheduler, "offset") WHERE consumer_scheduler = false;
+
+-- Go back to the original PK and queues_offset__type as they are useful for offset based poll and updates
+DO $$
+    BEGIN
+        IF NOT exists (select constraint_name from information_schema.table_constraints where table_name = 'queues' and constraint_type = 'PRIMARY KEY') then
+            ALTER TABLE queues ADD PRIMARY KEY("offset");
+        END IF;
+    END;
+$$;
+DROP INDEX IF EXISTS queues_offset;
+CREATE INDEX IF NOT EXISTS queues_type__offset ON queues (type, "offset");