Refactor midpoint function and fix many logical errors, overflow and functional bugs

[nihalzp]

Logical error:

SELECT midpoint(materialize(toNullable(1)), 11);

Wrong result for mixed numeric types:

SELECT midpoint(toUInt32(100), toDecimal32(200, 3));

Outputs 100.05 but should be 150.

Inconsistent rounding behavior:

SELECT midpoint(-4, -5); 
-- Returns -5 (rounds towards neg infinity)

SELECT midpoint(-10, -4, -2, -2); 
-- Returns -4 (rounds towards 0)

Overflow:

SELECT midpoint(toInt64('9223372036854775807'), toInt64('9223372036854775807'), toInt64('9223372036854775807'));

Returns 3074457345618258601, but should be 9223372036854775807.

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 into CHANGELOG.md):

Fix numerous logical errors, overflow and functional bugs in midpoint function. Closes #91816.

Documentation entry for user-facing changes

• Documentation is written (mandatory for new features)

[Copilot]

Pull request overview

This PR refactors the midpoint function to fix multiple critical bugs including incorrect handling of nullable types, mixed numeric type conversions, inconsistent rounding behavior, and integer overflow issues. The fix implements proper nullable-aware semantics that ignore NULL values per row and uses wider integer types during summation to prevent overflow.

Key changes:

• Reimplements nullable handling to correctly ignore NULL arguments per row instead of treating all rows as NULL
• Fixes integer overflow by using wider types (Int128/UInt128/Int256/UInt256) during summation before division
• Corrects rounding behavior for signed integers to consistently truncate toward zero

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
src/Functions/midpoint.h	Complete refactor of midpoint logic: adds nullable-aware calculation paths, overflow-safe summation using wider types, and consistent truncation-toward-zero rounding for integers
src/Functions/avg2.cpp	Simplifies avg2 (2-arg alias) by removing duplicate type resolution logic now handled by base class
tests/queries/0_stateless/03710_midpoint.sql	Adds comprehensive test coverage for nullable types, overflow cases, mixed types, and edge cases
tests/queries/0_stateless/03710_midpoint.reference	Expected results for new test cases
tests/queries/0_stateless/03710_midpoint_jit.sql	Adds JIT compilation test coverage for both positive and negative number cases
tests/queries/0_stateless/03710_midpoint_jit.reference	Expected results for JIT tests

[clickhouse-gh]

Workflow [PR], commit [ac5b006]

Summary: ⏳

job_name	test_name	status	info	comment
Integration tests (amd_asan, db disk, old analyzer, 4/6)		failure
	test_backup_restore_s3/test.py::test_backup_to_s3_different_credentials[data_file_name_from_first_file_name-native_multipart]	FAIL	cidb, issue
BuzzHouse (amd_debug)		failure
	Logical error: 'Inconsistent AST formatting in Function_tupleElement: the query: (STID: 1941-1bfa)	FAIL	cidb
BuzzHouse (amd_tsan)		failure
	Logical error: '(isConst() || isSparse() || isReplicated()) ? getDataType() == rhs.getDataType() : typeid(*this) == typeid(rhs)' (STID: 2508-3374)	FAIL	cidb
BuzzHouse (amd_ubsan)		failure
	Logical error: '(isConst() || isSparse() || isReplicated()) ? getDataType() == rhs.getDataType() : typeid(*this) == typeid(rhs)' (STID: 2508-2f8d)	FAIL	cidb

[nihalzp]

Integration tests (amd_asan, db disk, old analyzer, 4/6)

• Flaky test: test_backup_restore_s3/test.py::test_backup_to_s3_different_credentials[data_file_name_from_first_file_name-native_multipart] #90662

BuzzHouse (amd_debug)

• Logical error: 'Inconsistent AST formatting in Function_tupleElement: the query: (STID: 1941-1bfa) #92069
• Logical error: '(isConst() || isSparse() || isReplicated()) ? getDataType() == rhs.getDataType() : typeid(*this) == typeid(rhs)' (STID: 2508-35cc) #91710