refactor: drop unused exports detected by fallow auto-fix

[jrusso1020]

What

Run fallow fix --auto-fixable to drop the export keyword from symbols fallow identifies as unreachable from any entry point, restricted to cases where the symbol is still used internally in its own file.

Why

After #938 + #942 landed the config + CI audit gate, fallow reports 276 dead-code findings. The biggest mechanical chunk is unused exports — symbols exported "in case someone needs them" that nothing currently imports. Removing the export keyword makes them module-private, shrinks the public surface, and lets future cleanup catch genuinely-orphaned declarations.

How

npx fallow@2.75.0 fix --yes would have applied 165 de-exports. Inspecting the result against typecheck + lint + the fallow audit surfaced three problem categories:

False positives (4 files reverted):

• captureCost.ts — renderOrchestrator.ts has two separate import { ... } from "./render/captureCost.js" blocks. Fallow only saw the first and incorrectly flagged symbols from the second block as unused. (Real bug to file upstream.)
• propertyPanelHelpers.ts, domEditingLayers.ts — internal uses fallow's reachability missed.
• render.ts — exports consumed via await import("./render.js") dynamic imports in tests; fallow's static analysis doesn't follow these.

Cascading dead code (20 files reverted):
When fallow removed export from a symbol that wasn't used in-file either, oxlint immediately surfaces no-unused-vars. Those cases are properly cleaned up by deleting the declaration, not just dropping export. Doing that touches different code shapes (cascading unused imports, dead helpers) and is better as a separate, narrower PR.

Barrel cascades (4 more files reverted):
After force-rebasing onto main, fallow audit caught a third pattern: dropping a single re-export from a barrel file or a single export from a tightly-coupled file made the other symbols in that file become unused. Examples:

• Button.tsx — removing Button's and IconButton's exports orphaned the whole file
• renderOrchestrator.ts — removing one captureCost re-export left other dead re-exports
• telemetry/index.ts, portUtils.ts, ui/index.ts, hyperframeRuntimeLoader.ts — similar barrel-cleanup cascades

These need a "whole-file cleanup" pass, not a per-symbol auto-fix.

Kept the remaining 38 files where the de-export is purely an access-modifier change — symbol is still used internally, no follow-on findings.

Result

• Fallow dead-code findings: 276 → 208 (68 fewer unused exports)
• No behavior change — each symbol is still defined and used identically inside its file
• Zero lint / format / typecheck regressions
• Fallow audit passes on this PR (all remaining findings are inherited)

Test plan

• bun run --filter '*' typecheck — clean across all 8 packages
• bun run lint — 0 errors
• bun run format:check — clean
• npx fallow@2.75.0 audit --base origin/main --fail-on-issues — exit 0 (all 19 remaining changed-file findings are inherited)
• vitest: packages/cli 335/335, packages/core 917/917, packages/studio 576/576, packages/engine 605/605

Follow-ups

• Delete fully-dead declarations (the 20 files reverted) — properly clean up code where the export had no consumers internally either
• Barrel/cascade file cleanup — Button.tsx, renderOrchestrator.ts re-export barrel, telemetry/index.ts, ui/index.ts, etc. Each is its own small whole-file decision.
• File multi-import-block bug with fallow upstream
• Producer renderOrchestrator hub cycle — 8 circular deps, real refactor

[jrusso1020]

• refactor: drop unused exports detected by fallow auto-fix #944 👈 (View in Graphite)
• ci: add fallow audit job (PR-scoped, new-only gate) #942
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.