Add task cards for CI failures and Zig implementation completeness

[Copilot]

Created three task cards to address GitHub Actions test failures and ensure complete Zig implementation parity with the C/Behavior reference.

Task Cards Added

• TASK-057: Fix GitHub Actions test failures (HIGH)

  • zig-tests workflow failing consistently since 2025-12-14
  • Scoped to workflow config, Zig implementation, or test harness issues

• TASK-058: Ensure all Zig functionality fully implemented (HIGH)

  • Systematic audit against feature matrix (research/zig-cr/90-feature-matrix.md)
  • Known gap: crsql_fract_as_ordered currently stubbed (1/20 C oracle tests failing)
  • Covers replication surfaces, serialization, SQLite APIs, clock versioning, fractindex

• TASK-059: Design tests to expose unimplemented Zig functionality (MEDIUM)

  • 9 focus areas: edge cases, multi-table scenarios, merge conflicts, clock semantics, serialization, vtable ops, hook chaining, schema evolution, performance
  • Target: 5-10+ deterministic test scenarios beyond current 154/154 MVP coverage

Documentation Updates

Updated research/zig-cr/92-gap-backlog.md:

• Added "CI and Zig implementation health" section with cross-links to new tasks
• Note: Local tests passing (154/154 MVP, 52/52 parity) but CI red

All task cards follow AGENTS.md contract with bidirectional cross-links, acceptance criteria, and progress logs.

Original prompt

Add a new task card for us to unbreak GitHub action test failures

And another one to be sure to ensure all zig functionality is fully implemented.

And another to think about what additional tests would help expose unimplemented functionality in the zig implementation and close that gap

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.