[MattJackson]

TODO section D, items C13 and C10 (deletion is priority 1).

LOC (non-blank, non-comment .rs; tests = tests/ dirs, *_tests.rs, inline #[cfg(test)] mods)

	before (origin/dev)	after
production	266,609	266,609
test	416,146	416,145

This PR changes comments only, apart from one plugin-facing refusal string, so the LOC barely moves.

C13: markers

• Removed ~55 (TODO <item>) / 1.6.0-TODO … citations from comments in contract, kernel, root, plugin-loader, core-oauth2 and core-connector. The plane conformance refusal no longer prints (TODO L-R22).
• Removed the TRANSITIONAL, INTERIM, "until the spool lands" and owner-pointer markers. The work they describe is listed below for the TODO.
• Not markers, kept: the migrator's # TODO(migrate) output and its tests and docs (this is product behaviour), and plain-English "until its delta/refresh/sample lands" and "struck when its steps drop" (these describe runtime behaviour).
• Left for the PRs that delete the code: the markers in busbar-plane-llm (step 5), plugin-loader/store_v3.rs and store_adapter.rs (C5/C6), core-connector/src/test_support.rs and kernel governance/signing.rs (C7), and hooks/tests/m4_acceptance_suite.rs (C12 deletes it).

Remaining work, for the TODO and no longer in the code:

1. The data door still binds its pre-TLS SO_REUSEPORT listener (Listening::poll_accept_stream, main.rs per-core runtimes), and serve.rs's hyper SessionCaller/IngressCaller still serve. Both go when the framer's accepted::Caller serves the data door (K1 U6/U7, row 192 C root: the data door served through its framer's accept side, the internal switch, the proofs #1470).
2. The connector's own I/O thread (root/connector.rs::io_reactor) exists only because synchronous governance callers wait on the control runtime. It goes with that wait (D2/D3).
3. EgressTrust::judge_answer is still interim for the kernel's eight pooled-client builders. It goes when D1-D6 move onto conns (C2 / row 130 E5+E7: the kernel's outbound HTTP engine leaves busbar-kernel; the no-socket pin is empty #1230).
4. export.<sink>.durable: true is refused because core has no spool.
5. SigningKey::mint_audience is cfg(test) only until an OAuth AS mint path is a production caller.
6. plugin_log_file_boot stays #[ignore] (RED) until the scrape sink is a memory-ABI plugin_door! plugin.

C10: qa inventories

• qa/field-inventory.json is a dead ledger, deleted. Its generator (cargo xtask dialect wire) and its only reader (enumerate-cells.py) are gone, and nothing in busbar or plane-llm/mcp/a2a opens it. testing/shadow-oracle/cells.json names it only in a derived_from label string; that file is oracle data and is left untouched.
• qa/method-inventory.json is protection data, kept. Behaviour tests read it at the busbar pin: busbar-plane-mcp tests/method_inventory.rs and busbar-plane-a2a build.rs. Moving it into a busbar crate's tests would break those repos at their next pin bump. Its proper home is the plane-mcp/plane-a2a repos, which is listed for the TODO.