r2il: fix lossy v1 sink artifacts, add facts/residuals reader

[AdaWorldAPI]

What

Continues PR2 (.claude/plans/r2il-behavioral-ir-v1.md) after PR #101 shipped
the RefinedTruthSink trait + OfflineSink + the §12 corpus-profile finding.

While building the reader for sink.rs's TSV artifacts, found that the v1
writer (merged in #101) was write-only and lossy:

• The residuals TSV wrote only reason.as_str() plus the three facet
  coordinates — dropping ResidualReason's per-variant typed payload
  entirely, plus ResidualFact::at_prefix and ResidualFact::provenance.
• The facts TSV omitted FactProvenance.value (the value_id column).

No reader ever existed to catch this — v1 shipped without a round-trip test.

What changed

• Bumped FACTS_VERSION 1→2 (added value_id column) and
  RESIDUALS_VERSION 1→2 (widened 4→14 columns: typed reason payload, the
  FacetPrefix variant, and the full FactProvenance tuple) — additive per
  STAGED-CODEGEN-GUIDE.md §4 (append columns, never reorder, version the
  header).
• Added read_facts / read_residuals plus the shared codecs both facts and
  residuals need (provenance_cols/parse_provenance_cols,
  facet_cols/parse_facet_cols, prefix_cols/parse_prefix_cols,
  reason_payload_cols/parse_reason, concern_from_str,
  fact_kind_from_str, facet_from_raw).
• Added OpTag::parse — the exact inverse of as_str() (one arm per
  variant, None on anything unrecognized), needed so a residual reason's
  opcode payload can round-trip through the TSV. Named parse rather than
  from_str to avoid clippy's should_implement_trait lint against
  std::str::FromStr's canonical shape.
• New round-trip tests, both disable-run verified (mutated the write or
  parse path, confirmed red, restored, confirmed green):
  read_facts_round_trips_every_field_including_value_id,
  read_facts_refuses_an_unknown_opcode_rather_than_guessing,
  read_residuals_round_trips_a_two_opcode_payload_reason,
  read_residuals_round_trips_a_payload_free_reason_with_no_facet.

Not in this slice

Per PR2's own incremental scoping (see #101): DTO/codebook wiring beyond
what facet.rs already has, the round-trip reconstruction oracle (R2IL →
routes → semantic-equivalent R2IL — SPO is explicitly NOT this oracle per
the plan), and the optional SPO projection of semantic facts (calls/objects
only).

Test plan

• cargo test -p ruff_r2il — 40 lib + 12 integration tests, all green
• cargo clippy -p ruff_r2il --all-targets -- -D warnings — clean
• cargo fmt --check — clean
• Two disable-run verifications on the new round-trip tests (mutate →
  red → restore → green)
• uv run --only-group dev prek run --files <changed files> — clean

🤖 Generated with Claude Code

---

Generated by Claude Code

[cursor]

Bugbot couldn't run - usage limit reached

Bugbot is counted against Cursor usage for this user or team, and this run hit a usage or spend limit.

A user or team admin can review and increase usage limits in the Cursor dashboard.

(requestId: serverGenReqId_c20360c9-72bc-4fbb-a1c0-a3613c1aa113)

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, you can upgrade your account or add credits to your account and enable them for code reviews in your settings.