TRS P0a: investigate hierarchical protocol and Verilog parity

[nanavati]

First P0a review checkpoint — not a P0 pass

Stacked on design PR #177 (trs/hierarchy-first-design at 054ba3f6400bfed18c20c1552981212e8cc956bd). Executed with gpt-6-astra, reasoning effort ultra, as requested. Changes are isolated to src/trs/spike/p0-hierarchy/; production compiler/runtime files are unchanged.

Findings

• The existing ADynSched export is narrower than the required pinned Verilog-backend target. The conditional-call seed's regression explicitly expects successful Verilog compilation with G0100; the new initialized variant is authored but uncompiled.
• Hand-authored model checks contradict run-child-once and blanket pre-edge guard freezing for the supplied same-edge interaction. They do not establish general protocol sufficiency.
• Existing PosWrap demonstrates BSV above BVI, not an independently interpreted BSV child below opaque RTL. The proposed subordinate-port bridge needs review.

Deliverables

• Candidate public-port protocol, owner-local derivation table, invariants and explicit safety/progress/effect-order gaps.
• 79-entry inventory with 46 witness records and an explicitly open coverage census.
• Bounded Python experiments, initialized BSV fixture and intended pinned compiler/oracle recipe.
• Fail-closed runner, 21 harness/index tests, environment report and raw results.

Verification

Check	Result
Source identity	Pass: compiler/runtime subtrees match the pin; BVI reference available
Small model	Pass for supplied contracts only; 27 toy ladder configurations
Reporting/index tests	21 pass, including missing tools, overclaim rejection and optimized-Python rejection
BSC/Bluesim/RTL/Verilator/Rust/waveforms	Not run: required toolchains unavailable
Aggregate P0 status	Exit 2, blocked/inconclusive

Read RESULTS.md first, then PROTOCOL.md and inventory.json. Review the candidate, dynamic extraction gap and BVI bridge before substantial P0b implementation. Local public-contract completeness, admission/ME, exactly-once effects, compositional progress and real scaling remain open. No global schedule, private-descendant plan or scope relaxation is proposed.