PG19: fix runtime bugs hit outside the regression suite

[ihalatci]

Stacked PR (PR2 of the PG19 stack). Base = pg19-ci-test-matrices (#8616), which is itself stacked on pg19-ruleutils-port (#8602) → pg19-support. Review/merge in stack order.

Fixes two PG19 runtime crashes that surface in normal Citus operation (outside the regression suite):

1. FuncnameGetCandidates() gained an int *fgc_flags out-parameter that the callee writes to unconditionally. The compat shim passed NULL and crashed on every by-name function lookup (e.g. the maintenance daemon). Now passes the address of an int compound literal.
2. TupleDesc->firstNonCachedOffsetAttr: PG19 added an offset cache that BlessTupleDesc() / slot_deform_heap_tuple() now assert is populated but no longer populate themselves. BlessTupleDesc is wrapped to call TupleDescFinalize() first (no-op shim on older majors), with explicit calls added at the four sites that deform hand-built TupleDescs.

Also installs ApplicationNameAssignHook via union access on PG19.

Closes #8610
Part of #8597

[codecov]

Codecov Report

❌ Patch coverage is 62.50000% with 3 lines in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (pg19-support@87fe895). Learn more about missing BASE report.

Additional details and impacted files

@@               Coverage Diff               @@
##             pg19-support    #8617   +/-   ##
===============================================
  Coverage                ?   88.93%           
===============================================
  Files                   ?      288           
  Lines                   ?    64371           
  Branches                ?     8092           
===============================================
  Hits                    ?    57251           
  Misses                  ?     4783           
  Partials                ?     2337           

🚀 New features to boost your workflow:

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

[Copilot]

Pull request overview

This PR is part of the PG19 support stack and addresses PG19-only runtime crashes by updating Citus compatibility shims and ensuring newly-required tuple descriptor finalization is performed before slot deformation.

Changes:

• Fix PG19 FuncnameGetCandidates() compatibility shim by providing a valid fgc_flags out-parameter.
• Ensure PG19 TupleDesc offset-cache initialization via TupleDescFinalize() (including a PG19 BlessTupleDesc() wrapper and explicit finalize calls at key call sites).
• Update GUC override logic for PG19’s config_generic layout change when installing the application_name assign hook.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
src/include/pg_version_compat.h	Updates PG19 compatibility macros (fgc_flags shim; TupleDescFinalize + BlessTupleDesc wrapper; no-op TupleDescFinalize on older majors).
src/backend/distributed/stats/query_stats.c	Finalizes SRF-provided TupleDesc before using it with slot deformation on PG19.
src/backend/distributed/shared_library_init.c	Adjusts PG19 GUC struct access to install ApplicationNameAssignHook safely.
src/backend/distributed/planner/multi_explain.c	Finalizes a hand-built TupleDesc used for EXPLAIN ANALYZE result collection on PG19.
src/backend/distributed/executor/distributed_intermediate_results.c	Finalizes hand-built TupleDescs used with tuple deformation/processing on PG19.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[ihalatci]

Good catch. Replaced the comma-expression macro with a single-evaluation static inline citus_BlessTupleDesc() helper defined before the macro, so the argument is only evaluated once and side-effecting call sites like BlessTupleDesc(CreateTemplateTupleDesc(...)) are safe. Verified with a full backend build against PG19beta1 (no warnings/errors).

[alperkocatas]

Nit: For item 2: The message says "call it explicitly at the three sites," but there are four explicit TupleDescFinalize() calls (two in distributed_intermediate_results.c, one in multi_explain.c, one in query_stats.c). The query_stats.c one is arguably a separate "defensive SRF" case rather than a hand-built desc — worth tightening the wording so the count matches.