fix(rego): schedule comprehensions in rule outputs

[Maksym (maksym-mishchenko)]

Problem

An inline comprehension in a guarded rule output loads and compiles, but interpreter evaluation fails with statements not scheduled in query {query:?} when the guard succeeds. Moving the comprehension into a separate rule works; RVM already handles the inline form.

The reported empty-input failure did not reproduce: empty or missing collections correctly leave the guarded rule undefined, not false.

Fix

I schedule comprehensions in output expressions with the body's local variables visible, including nested outputs, function results, and bodyless partial-set keys. Branch analysis follows the output actually evaluated, preserving else and independent partial-rule body semantics.

Regression tests cover interpreter/RVM parity and fresh-engine eval_rule/eval_query, including empty and missing inputs and duplicate/false array elements.

Verification

• 24 comprehension cases, 29 Rego YAML test functions, and the public Engine suite passed.
• cargo xtask ci-debug passed.
• Repository pre-commit and pre-push checks passed: build, workspace/binding formatting and Clippy, doc tests, no_std build, and OPA conformance (2,875 passed, 0 failed).

[Copilot]

Copilot review overview

🟢 Approval recommended

The scheduler logic aligns with interpreter branch semantics and is comprehensively covered by focused regression tests.

Review effort: Balanced
Findings: None

What changed in this PR

Schedules inline comprehensions in Rego rule outputs so interpreter evaluation matches RVM behavior.

Changes:

• Schedules comprehensions using body-local scope.
• Handles nested, function, partial-rule, and else outputs.
• Adds interpreter/RVM parity and fresh-engine regression tests.

File	Description
src/​scheduler.rs	Adds output-expression scheduling and nested-scope propagation.
tests/​engine/​mod.rs	Tests fresh-engine evaluation and variable safety.
tests/​rvm/​rego/​cases/​comprehensions.yaml	Covers comprehension output variants and parity.

---

💡 Configure MCP servers for context-aware, tailored reviews. Learn more in the docs.