test: repr_py control-character conformance tests

[leongdl]

Two fixtures pinning Expression Language §2.2.6's requirement that repr_py escape control characters, promoted out of proposed/ now that the fix has landed.

TLDR. Both were parked here as expected failures. openjd-rs#374 merged as 5b04959, so both pass and become the regression tests the proposed/README.md promised. Verified against a pre-fix binary as the control: both fail before the fix and pass after, so they pin what they claim.

The two fixtures

Fixture	What it pins
EXPR/jobs/expr2.2.6--repr-py-newline-roundtrip	repr_py("a\nb") fed to ast.literal_eval round-trips to the input
WRAP_ACTIONS/jobs/wrap-repr-py-escapes-newline-in-wrapped-args	the same rule through RFC 0008's repr_py(WrappedAction.Args) forwarding pattern, with a two-line python -c body

The pair is deliberate. The EXPR one isolates the rule; the WRAP_ACTIONS one exercises the end-to-end path a template author actually writes, where the value crosses two process boundaries before being parsed.

Measured

Built openjd from openjd-rs main at 5b04959, and separately from a pre-fix commit as the control:

Fixture	pre-fix	merged
expr2.2.6--repr-py-newline-roundtrip	FAIL	PASS
wrap-repr-py-escapes-newline-in-wrapped-args	FAIL	PASS

Both pre-fix failures are SyntaxError: unterminated string literal with the expected output missing, which is exactly the defect: repr_py quoted its output without escaping, so a value carrying a newline emitted it raw and CPython would not parse the result.

Run through the runner's directory scan from the new locations, not single-file mode, since discoverability is the point of leaving proposed/:

2023-09/EXPR/jobs:
  ✓ expr2.2.6--repr-py-newline-roundtrip
2023-09/WRAP_ACTIONS/jobs:
  ✓ wrap-repr-py-escapes-newline-in-wrapped-args

No regression in either category: WRAP_ACTIONS 73 passed, 0 failed and EXPR 352 passed, 0 failed.

Notes for review

The headers are rewritten rather than moved unchanged. Both described the parked state and asserted that the implementations fail, which is no longer true. They now carry spec, test and expectation in three paragraphs, name the paired fixture, and sit inside the comment-line budget.

proposed/README.md and the proposed/ directory are removed. The fixture it described was its only entry.

The EXPR fixture originates on the func-lib expected-failures branch behind #165, which should drop its copy so the two do not collide on merge.

One thing this does not resolve, recorded so it is not lost: Template Schemas §5.2 restricts <ArgString> to characters outside the Cc category, which excludes the newlines both of these fixtures rely on. Neither implementation enforces that, and 150 of 1,191 merged fixtures carry a newline in an args entry, so the multi-line python -c convention is load-bearing suite-wide. Either §5.2 is relaxed to permit U+000A and U+000D, or that convention is invalid. That is a spec decision and a separate PR; these fixtures are correct under the reading both implementations already follow.

---

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[leongdl]

Quorum verdict: GOOD-WITH-NITS (4 reviewers flagged variants of the same two issues) — spec: EL §2.2.6 repr_py (newline must be escaped — spec gives the exact worked example) via RFC 0008 WrappedAction.Args forwarding.

(1) Spec self-tension, correctly disclosed but under-weighted: the fixture's own onRun args contain a literal U+000A, which §5.2's ArgString Cc-exclusion arguably forbids — as written the template may be spec-invalid, so promotion is contingent on resolving §5.2 vs the multi-line python -c convention (which the Deadline Docker container env also relies on; if §5.2 is enforced, that breaks service-wide). Resolve the spec question first.
(2) Attribution: README says "implementation bug (openjd-rs)" but the 2026-08-12 sweep shows it fails BOTH implementations — state Python's status.
(3) Partial overlap with PR #165's expr2.2.6--repr-py-newline-roundtrip (same root defect); the end-to-end wrap path justifies keeping both, but cross-reference them.

[leongdl]

Quorum review (5 independent agents). 1 fixture + README. Verdict: GOOD-WITH-NITS. The repr_py newline defect is real (fails BOTH implementations per the 2026-08-12 sweep) and the end-to-end wrap-forwarding angle adds value over PR #165's unit-level twin — but the fixture is hostage to an unresolved spec tension its own README flags, and the README under-attributes the failure. Cross-PR: this branch uses component-level WRAP_ACTIONS/proposed/ while PR #166 uses kind-level job_templates/proposed/ — the five expected-failures PRs should standardize on one placement (kind-level routes promotion mechanically).

[leongdl]

Quorum-review fixes applied and pushed (rebased onto mainline 3069673):

• Restructured to kind-level parking: WRAP_ACTIONS/jobs/proposed/ (promotion = move up one directory).
• README corrected: the repr_py newline defect fails both implementations (2026-08-12 sweep), not just openjd-rs.
• The §5.2 ArgString/newline conflict is now the stated promotion gate rather than a footnote — as written the fixture's own args are arguably spec-invalid until that's resolved (and the resolution also decides the fate of the multi-line python -c convention suite-wide).
• Cross-referenced the unit-level twin in the func-lib expected-failures PR (test: [expected failures] function-library fixtures pending spec decisions #165).