refactor: kill J namespace barrel (Phase 1, #234)

[pajoma]

Phase 1 of #234 — package-by-feature restructure.

Removes the J namespace barrel that created the root import cycle (src/index.ts → submodule → src/index.ts).

Changes

• Replace import * as J from '<root>' + J.<Sub>.<Sym> with named imports from each submodule barrel across 47 files.
• Also fixed a non-codemod variant: import { Util } from '..' in vscode/conf.ts.
• Empty src/index.ts (export {}) to break the cycle.

Why behavior-identical

J.<Sub>.X already resolved through src/<sub>/index.ts. Rewriting to named imports from the same barrel changes resolution path only by dropping the root re-export — no symbol resolves differently.

Verification

• npm run compile-tests (tsc) ✔
• npm run compile (esbuild) ✔
• npm run lint ✔
• Full suite: 238 passing, exit 0

Notes

• Pure mechanical, zero behavior change. Largest diff of the restructure — unblocks Phases 2–5.
• Acceptance: no import * as J remains; root barrel reduced to export {}.

Refs #234.

🤖 Generated with Claude Code

[pajoma]

Brooks-Lint PR Review: Approved

Health Score: 95/100

Phase 1 of #234 is successfully implemented. By replacing the global \J\ namespace barrel with granular named imports, this PR breaks the primary circular dependency cycle that previously coupled the entire system.

Key Findings:

• Architecture: The root import cycle (\src/index.ts\ → submodule → \src/index.ts) is now broken. This enables better tree-shaking and clear dependency tracking.
• Verification: Purely mechanical change with zero behavior shift. Confirmed by 238 passing tests.
• Risk: High file churn (47 files) is necessary and expected for this structural fix.

Recommendation: Merge immediately to unblock Phase 2 (DI Container) and minimize rebase conflicts.