feat: AI Agent-driven test infrastructure (Phase 1)

[Copilot]

Foundational infrastructure for automated test planning from PR diffs: parse changed files → map to test categories → generate structured TestPlan → deduplicate SQL cases → execute via local runner. Phase 1 of agent-driven CI.

New packages (pkg/testinfra/)

• types/ — Domain types: TestPlan, TestTask, DiffSummary, FileChange, priority/status enums, JSON round-trip
• planner/ — Unified diff parser (files, packages, functions from hunk headers), 22 static path→test-category mapping rules (e.g. pkg/sql/plan/ → optimizer BVT, pkg/vm/engine/disttae/ → disttae UT), Planner combining diff + mapping into a prioritized TestPlan with SCA/UT/BVT tasks
• executor/ — Executor interface + LocalExecutor wrapping go test/go vet/mo-tester, sequential ExecutePlan orchestrator
• dedup/ — SQL fingerprinting (normalize case/whitespace/literals → SHA-256), .test file parser handling mo-tester tags (@bvt, @sortkey), batch dedup with duplicate-group reporting

CLI + CI

• cmd/mo-testplan/ — git diff | mo-testplan --pr 42 --base main --format json|summary
• .github/workflows/testplan.yaml — PR open/sync trigger → generate TestPlan → upload artifact → post summary comment

Usage

git diff origin/main...HEAD | go run ./cmd/mo-testplan --pr 42 --base main --head feature/x --format summary
# TestPlan for PR #42
# Files changed: 5, Tasks: 15 (UT: 2, BVT: 12, SCA: 1)

go test ./pkg/testinfra/...   # all 4 packages pass

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}