Fixes #27158: tag_usage Postgres seq-scan (1.12.7 backport) (#27824)

* Fixes #27158: restore tag_usage prefix-LIKE index on Postgres

The 1.11.0 perf migration (#23054) added four `WHERE state = 1` partial
indexes on tag_usage; #24063 dropped the matching `state = 1` predicate
from getTagsInternalByPrefix (Suggested-state rows are valid for both
classification and glossary derivation), leaving every partial index
inapplicable. Postgres fell back to a parallel seq scan; MySQL was
unaffected because its 1.11.0 indexes were never partial.

Adds non-partial single-col btrees on targetfqnhash_lower and
tagfqn_lower (mirror MySQL's idx_targetfqnhash_lower / idx_tagfqn_lower)
and rebuilds the four partials as non-partial -- same shape, same
INCLUDE columns, predicate coupling removed so future query changes
can't silently invalidate them.

Backport of #27745 (main) onto the 1.12.7 release line so customers on
1.12.x get the fix without waiting for 2.0.x.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

* Drop new indexes before CREATE to defuse failed-CONCURRENTLY edge case

If CREATE INDEX CONCURRENTLY fails partway (lock timeout, OOM,
connection drop on a busy multi-GB tag_usage), Postgres leaves the
index in an INVALID state. A subsequent CREATE ... IF NOT EXISTS sees
the catalog row and silently skips, leaving the index permanently
broken while the migration reports success.

The four composite indexes already use the DROP-then-CREATE pattern;
applying the same to the two new single-col indexes for symmetry.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: sonika-shah

bootstrap/sql/migrations/native/1.12.7/mysql/postDataMigrationSQLScript.sql

@@ -0,0 +1 @@
+-- Placeholder for 1.12.7 MySQL post-data migration script

bootstrap/sql/migrations/native/1.12.7/mysql/schemaChanges.sql

@@ -0,0 +1,4 @@
+-- Placeholder for 1.12.7 MySQL schema changes
+-- The Postgres-side fix for #27158 has no MySQL counterpart: MySQL's
+-- 1.11.0 indexes were already non-partial (no partial-index syntax in
+-- MySQL), so the regression that hit Postgres did not affect MySQL.

bootstrap/sql/migrations/native/1.12.7/postgres/postDataMigrationSQLScript.sql

@@ -0,0 +1 @@
+-- Placeholder for 1.12.7 PostgreSQL post-data migration script

bootstrap/sql/migrations/native/1.12.7/postgres/schemaChanges.sql

@@ -0,0 +1,32 @@
+-- Issue #27158: tag_usage seq-scan on Postgres. #24063 dropped the
+-- `state = 1` predicate that 1.11.0's partial indexes required.
+-- Fix: add single-col indexes on the `_lower` columns, and drop the
+-- `WHERE state = 1` filter from the partials so changes can't invalidate them.
+
+DROP INDEX CONCURRENTLY IF EXISTS idx_tag_usage_targetfqnhash_lower_pattern;
+CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_tag_usage_targetfqnhash_lower_pattern
+ON tag_usage (targetfqnhash_lower text_pattern_ops);
+
+DROP INDEX CONCURRENTLY IF EXISTS idx_tag_usage_tagfqn_lower_pattern;
+CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_tag_usage_tagfqn_lower_pattern
+ON tag_usage (tagfqn_lower text_pattern_ops);
+
+DROP INDEX CONCURRENTLY IF EXISTS idx_tag_usage_target_prefix_covering;
+CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_tag_usage_target_prefix_covering
+ON tag_usage (source, targetfqnhash_lower text_pattern_ops)
+INCLUDE (tagFQN, labelType, state);
+
+DROP INDEX CONCURRENTLY IF EXISTS idx_tag_usage_tagfqn_prefix_covering;
+CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_tag_usage_tagfqn_prefix_covering
+ON tag_usage (source, tagfqn_lower text_pattern_ops)
+INCLUDE (targetFQNHash, labelType, state);
+
+DROP INDEX CONCURRENTLY IF EXISTS idx_tag_usage_join_source;
+CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_tag_usage_join_source
+ON tag_usage (tagFQNHash, source)
+INCLUDE (targetFQNHash, tagFQN, labelType, state);
+
+CREATE EXTENSION IF NOT EXISTS pg_trgm;
+DROP INDEX CONCURRENTLY IF EXISTS gin_tag_usage_targetfqn_trgm;
+CREATE INDEX CONCURRENTLY IF NOT EXISTS gin_tag_usage_targetfqn_trgm
+ON tag_usage USING GIN (targetFQNHash gin_trgm_ops);