feat(analysis): authoritative network-wide NC bound = min(PLP,TFA) (REQ-NC-PLP-MIN-001)

[avrabe]

First v0.22.0 feature ("Deployable TSN synthesis" theme). Closes the V for REQ-NC-PLP-MIN-001.

What

network_wide_nc_bounds now emits a per-stream WcttAuthoritativeBound = min(TFA, PLP) — the tightest sound per-flow network-calculus bound. PLP and TFA are incomparable sound over-approximations (both ≥ exact, neither dominates), so their pointwise minimum is the number downstream consumers should trust: never worse than TFA, with the PLP win wherever PLP is tighter.

• Seeds the authoritative vector from TFA (always computed); the milp-solver PLP arm tightens it via min.
• When PLP is absent (feature off, or a non-feed-forward topology) the authoritative bound is TFA alone.
• min over two sound bounds is itself sound → no new soundness surface; this is pure presentation/selection and does not change either engine.

Oracle — TEST-NC-PLP-MIN

network_wide_authoritative_bound_is_min_of_plp_and_tfa runs the converging-tree fixture (bridge_converge_aadl) — the documented incomparability counterexample where the fast stream's pure PLP (~1148µs) exceeds its TFA (~1048µs) — and asserts:

1. one authoritative bound per stream, emitted regardless of the milp-solver feature;
2. with milp-solver, authoritative == min(PLP, TFA) for every stream;
3. on the PLP>TFA cross-burst stream the authoritative bound equals TFA and sits strictly below PLP (a regression that blindly reported PLP fails here);
4. without milp-solver, the authoritative bound equals TFA.

Genuinely red→green (WcttAuthoritativeBound did not exist before).

Gates (local)

• 893 spar-analysis lib tests pass (milp-solver); no-default-features path compiles + passes.
• clippy -D warnings clean; nightly fmt clean.
• mutation --in-diff: 1/1 caught (the .min mutant is killed by the counterexample oracle).

Falsification

If WcttAuthoritativeBound ever exceeds either the TFA or the PLP bound for a stream, or omits a stream that has a TFA bound, the selection is wrong.

🤖 Generated with Claude Code

[codecov]

Codecov Report

❌ Patch coverage is 90.47619% with 4 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
crates/spar-analysis/src/wctt.rs	90.47%	4 Missing ⚠️

📢 Thoughts on this report? Let us know!

[github-actions]

Rivet verification gate

✅ 20/20 passed

	count
Passed	20
Failed	0
Skipped (no steps)	0

Filter: (and (= type "feature") (or (has-tag "v093") (has-tag "v0100")))

Failed artifacts

(none)

_{Updated automatically by tools/post_verification_comment.py. Source of truth: artifacts/verification.yaml.}