src/v2/compiler audit

[gunbai-bot]

Auto-opened by session-dashboard for session bold-bat-503.
Pushing to session/bold-bat-503 advances this PR.

Worker attestation

Before flipping this PR to ready for review, confirm each item:

• Title describes the change (not the session id or branch).
• PR body summarises what and why (replace the TODO below).
• Tests run: name the command (e.g. npm test, cargo test) and the result.
• If this closes a work item, the body contains a Closes #N directive.
• No commits on this branch are surprises (no fork/cherry-pick I did not make).
• No secrets / credentials / large binaries staged.

Summary

TODO: replace this paragraph with one or two sentences naming the change and its motivation. Reviewers read this first.

Test plan

• TODO: list the commands that ran (or "no tests changed; relied on CI") and the outcome.

[gunbai-bot]

CI failures here are pre-existing main-floor breakage, not caused by this PR.

The 4 failing checks all stem from product.compute_fabric no longer exporting names that dsl/gunbc/* still import (ComputeOffer, compute_host_ram_bytes_total, ResourceEnvelope, MemoryRequirement, …). These were removed/renamed by the #5889 compute_fabric re-model that just landed on main. Verified:

• main's own CI runs are failing on the same errors (the Re-model compute_fabric as a namespace need<->opportunity connector; shape DERIVED from the .dag program, no heavy/workload/locality (replaces discarded #5866) #5889 run and every run after it are red).
• ComputeOffer/compute_host_ram_bytes_total are absent from compute_fabric on origin/main.
• This PR's entire diff is a 2-line identifier rename in src/v2/compiler/06_translate.dag (an internal helper, no call sites outside the file) — it touches nothing in dsl/gunbc/ or compute_fabric.

The fix belongs to the #5889 compute_fabric re-model owner (reconcile the dsl/gunbc/* consumers to the new connector shape), not to this rename. Once main is green again, this branch will go green with no changes.

— sent from bold-bat-503

[gunbai-bot]

Update: the single remaining CI failure is also pre-existing main-floor breakage, not from this PR.

The failing witness is ✗ FAIL [batch 4] regen_verify_gate_passes (returned Bool(false)) — the two-generation regen self-host fixed-point gate (#5873).

Proof it is a main regression, not this rename:

• main's own revert commit Revert #5889 (compute_fabric re-model) to restore main green #5901 (106c3978) fails CI on the identical witness — ✗ FAIL [batch 4] regen_verify_gate_passes — with no other errors.
• Earlier main runs masked it: they failed at batch 1 (compile-clean) on the Re-model compute_fabric as a namespace need<->opportunity connector; shape DERIVED from the .dag program, no heavy/workload/locality (replaces discarded #5866) #5889 compute_fabric errors and stopped before batch 4 ran. Now that Revert #5889 (compute_fabric re-model) to restore main green #5901 reverted those compile errors, batch 4 is reached and the regen gate fails on main.
• This PR's diff is a 2-line internal rename in 06_translate.dag; the symbol appears nowhere in the committed seed (src/v1/**), so it cannot shift the regenerated-vs-committed seed comparison the gate performs.

The fix belongs to the regen fixed-point owner (#5873 area — re-sync the seed to the fixed point). This branch will go green once main's regen gate is green again; no change is warranted here.

— sent from bold-bat-503