Backport of #93814 to 25.8: Fix logical error in Map column types when used with empty tuple

[ianton-ru]

Backport of ClickHouse#93814 by @nihalzp

Changelog category (leave one):

• Critical Bug Fix (crash, data loss, RBAC) or LOGICAL_ERROR

Changelog entry (a user-readable short description of the changes that goes to CHANGELOG.md):

Fix logical error in a rare case when empty tuple is used with Map column. Closes ClickHouse#93784. (ClickHouse#93814 by @nihalzp)

CI/CD Options

Exclude tests:

• Fast test
• Integration Tests
• Stateless tests
• Stateful tests
• Performance tests
• Aarch64 tests
• All with ASAN
• All with TSAN
• All with MSAN
• All with UBSAN
• All with Coverage
• All Regression
• Disable CI Cache

Regression jobs to run:

• Fast suites (mostly <1h)
• Aggregate Functions (2h)
• Alter (1.5h)
• Benchmark (30m)
• ClickHouse Keeper (1h)
• Iceberg (2h)
• LDAP (1h)
• OAuth (5m)
• Parquet (1.5h)
• RBAC (1.5h)
• SSL Server (1h)
• S3 (2h)
• S3 Export (2h)
• Swarms (30m)
• Tiered Storage (2h)

[github-actions]

Workflow [PR], commit [cba9643]

[ianton-ru]

Failed stateless tests looks unrelated

03707_set_index_bad_get_null_bug - looks like need changes for explain query plan from ClickHouse#105036

03579_create_table_populate_from_s3 - known bug ClickHouse#107057

[mkmkme]

Identical to the upstream one. LGTM

[CarlosFelipeOR]

✅ QA Verification — PASS

Verdict

No regressions introduced by this PR. All CI failures are either infrastructure flakiness (Keeper) or already-resolved issues (#2008). Approved from QA side.

Summary

Backport of upstream #93814 to stable-25.8. The fix (src/Functions/array/array.cpp: insertManyDefaults(columns.size() * input_rows_count)) correctly resolves the LOGICAL_ERROR on empty Tuple() inside a Map column with multiple rows.

The new test 03786_empty_tuple_map_array passed in the following CI jobs:

• Stateless tests (amd_binary, ParallelReplicas, s3 storage, parallel)
• Stateless tests (amd_binary, old analyzer, s3 storage, DatabaseReplicated, parallel)
• Stateless tests (amd_debug, AsyncInsert, s3 storage, parallel)
• Stateless tests (amd_debug, distributed plan, s3 storage, parallel)
• Stateless tests (amd_debug, parallel)

CI failures — all justified, none is a regression

1. Server died in Stateless tests (amd_binary, old analyzer, s3 storage, DatabaseReplicated, sequential) — Infrastructure (Keeper session-loss cascade), unrelated to this PR.

• The FAIL rows are crash meta-statuses (Server died, Killed by signal, Fatal messages), not test assertions. The test tagged SERVER_DIED was 03532_crash_in_aggregation_because_of_lost_exception, with a ~36 min duration — it hung, it did not fail.
• Root cause from the logs: at ~14:25:58 all three DatabaseReplicated nodes lost their Keeper session (Nothing is received in session timeout of 30000 ms, KEEPER_EXCEPTION; 269 occurrences on server1). ~58 min later the CI watchdog killed the servers with signal 6. The crash stack is entirely in shutdown (DDLWorker::shutdown() → ThreadFromGlobalPool::join() → Poco::Event::wait()), i.e. the DDL worker hung because it could not reach Keeper.
• 03532 passes on every other run, including this exact job on other PRs and on master. This PR only touches FunctionArray (array.cpp) — nothing related to Keeper, aggregation, or DatabaseReplicated.

2. 03707_set_index_bad_get_null_bug — Resolved by the fix backported in #2008.

3. 03579_create_table_populate_from_s3 — Marked as broken in tests/broken_tests.yaml via #2008.

Both #2008 changes are already merged into stable-25.8 (the base branch of this PR), so these two failures are addressed once this branch is up to date.