Emit use lines for dealiased grounding types (E0425=140 class)

[gunbai-bot]

Summary

Fresh-emit residue after #6243/#6295 was E0425≈140 (Magnitude×73 + CommutativeSemiring×55 + Nat×2): grounding aliases like Nat = CommutativeSemiring<Magnitude> emitted foreign RHS leaf types unqualified while the import graph suppressed them as non-emittable opaque decls.

• Supplemental use-line emission in emit_module_full: walk type-alias RHS trees, collect foreign leaf names that render unqualified, and emit use crate::<mod>::<Type>; lines merged into the import block.
• Authored-alias preservation extended from fn-sig/data-def (Emitter restoration: deref boxing + alias-brand rendering (C8 tail) #6243) to render_rust_decl_type and struct field types so imported Nat stays Nat instead of peeling to the dealiased RHS.
• Stage0 emitter synced; discriminating witnesses added.

Test plan

• cargo test -p v1-compiler-tests dealiased_grounding -- --nocapture — 2 passed
• cargo test -p v1-compiler-tests data_def_brand_alias_accessor_preserves_declared_type -- --nocapture — passed (regression)
• CI floor (auto on ready)

[gunbai-bot]

Re: review #35854 — the runner_slot_allocation.dag hunk is no longer in this PR. After #6309 landed on main, I rebased onto d48d5b1 and dropped commit c98fbb1 (rebase --skip); HEAD 385e1db diff is emit-only (4 files: 05_emit_rust.dag, stage0 mirror, witness test, lib.rs #[cfg(test)]). The approval's substance on the emitter fix stands; no further code changes requested.

— sent from quick-koi-412

[gunbai-bot]

Re: review #35869 — that approval reviewed an intermediate broken rebase (c9d6183) where a botched git rebase origin/main accidentally dropped the emitter hunks, leaving only the two witness tests in the diff. Fixed and force-pushed as a6a83d696f: full 4-file diff restored (05_emit_rust.dag, stage0 mirror, witnesses, #[cfg(test)] gate) on top of main @ 146d7e55b8 (#6310 #6304-revert). The approval's substance on the witnesses stands; the emitter fix is back in scope for re-review on a6a83d.

— sent from quick-koi-412

[gunbai-bot]

Re: review #35877 — agree the cli_run.rs hunk is not emitter logic; it landed here because CI blocked merge on a6a83d: touching src/v1/05_emit_rust.dag put that path in the affected-set node frontier, and floor_diff_edits_from_line_ranges tried to resolve it against witness-layer roots (dag + src/v2 only) → discovery-corpus fail-closed on unresolved v1.compiler.* imports. path_under_witness_layer_root skip matches ci_layer_roots.dag ("src/v1 excluded from witness_layer_roots"). Happy to peel to a follow-up PR if operator prefers, but it was a receipt prerequisite for this emit change, not scope creep.

Emitter substance verified on cfa2e21: supplemental use-lines + authored-alias preservation + witnesses unchanged.

— sent from quick-koi-412

[gunbai-bot]

CI @ cfa2e21 diagnosis — not a regression from this PR:

Check	Result
rust_tests	✅ green
floor batch 1 (dag_compile_clean)	✅
floor batch 2 (discovery-corpus, layering, …)	✅ — path_under_witness_layer_root fix confirmed
floor batch 3 source_root_ingest_gate_passes	❌ Bool(false)

Same failure on main @ 146d7e55b8 (run before this PR's floor): identical ✗ FAIL [batch 3] source_root_ingest_gate_passes. Main has been red on the floor since ≥#6304; #6310 did not clear it.

This PR's diff is src/v1 only (emit + witnesses + witness-root scoping). source_root_ingest_gate exercises src/v2/workflow/source_root_ingest_transport.dag shell orchestration — out of scope and unchanged vs main.

PR-local bar met: rust_tests green + discovery-corpus green. Lane blocked on main-wide SourceRootIngestGate red — needs operator/main fix, not another emit commit here.

— sent from quick-koi-412