PG-2014 Initalize nested query stack to fix crash on DDL

[jeltz]

The pgsm_ProcessUtility() which handles DDL increments nesting_level but does not put a query text on the next_queries stack while pgsm_ExecutorRun() does both. It is unclear to me if this is a mistake or by design but since readers of the query check for is the query text pointer is NULL and pgsm_ExecutorRun() reset the pointer to NULL before returning it is safe as long as we initialize the stack to all NULL pointers, which we did not.

This bug was found by our test suite in Jenkins on some RHEL based distro version and seems to mostly happen when the first query of a backend is CREATE EXTENSION and we have enabled query normalization but it is entirely possible that it could happen under other circumstances too.

The use of calloc() over palloc0() is to keep the patch small since the previous code used malloc().

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 85.48%. Comparing base (b00caaf) to head (3a9975d).
⚠️ Report is 3 commits behind head on release-2.3.1.

Additional details and impacted files

@@                Coverage Diff                @@
##           release-2.3.1     #584      +/-   ##
=================================================
+ Coverage          85.46%   85.48%   +0.01%     
=================================================
  Files                  3        3              
  Lines               1342     1343       +1     
  Branches             215      215              
=================================================
+ Hits                1147     1148       +1     
  Misses                92       92              
  Partials             103      103              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.