Fix npm global prefix fallback

[konard]

Fixes #54

Summary

• Preflight the npm global root before npm-backed installs and redirect installs to a use-m-owned cache prefix when the configured root is not writable.
• Preserve explicit npm_config_prefix / NPM_CONFIG_PREFIX choices instead of overriding them, with an actionable error if the configured root is not writable.
• Apply the npm resolver change across use.mjs, use.cjs, and use.js, plus a sync regression test to keep those distributed entrypoints aligned.
• Refresh Deno remote integrity and harden current TypeScript/Bun-related tests so CI matches the current runtime/toolchain behavior.

Reproduction And Verification

• Added fake-npm regression tests for a non-writable global root, a writable default root, and an explicit configured prefix.
• The non-writable case simulates npm root -g under /sys/...; the resolver now installs into ${XDG_CACHE_HOME}/use-m/npm-global by passing npm_config_prefix to npm.

Tests

• npm test -- --runInBand
• bun test
• deno test --allow-net --allow-env --allow-run --allow-read --allow-write --allow-sys

[konard]

Working session summary

Implemented and pushed the fix to PR 55:

#55

What changed:

• npm resolver now checks npm root -g before install and falls back to ${XDG_CACHE_HOME}/use-m/npm-global / ~/.cache/use-m/npm-global when the global root is not writable.
• Explicit npm_config_prefix / NPM_CONFIG_PREFIX is respected.
• use.mjs, use.cjs, and use.js were updated together.
• Added regression tests for non-writable global roots, writable roots, explicit prefixes, and a sync guard for the three distributed scripts.
• Fixed current CI drift in TypeScript expectations, Bun runtime detection contamination, and Deno lock integrity.

Verification:

• npm test -- --runInBand
• bun test
• deno test --allow-net --allow-env --allow-run --allow-read --allow-write --allow-sys
• Fresh GitHub Tests workflow: passed
• Fresh GitHub CI/CD workflow: passed

PR 55 is marked ready for review, merge state is clean, and the local worktree is clean at 30e5049.

---

This summary was automatically extracted from the AI working session output.

[konard]

🤖 Solution Draft Log

This log file contains the complete execution trace of the AI solution draft process.

💰 Cost estimation:

• Model: GPT-5.5
• Provider: OpenAI
• Public pricing estimate: $12.431564

📊 Context and tokens usage:

• 296.7K / 1.1M (28%) input tokens, 46.6K / 128K (36%) output tokens

Total: (296.7K + 7.4M cached) input tokens, 46.6K output tokens, $12.431564 cost

🤖 Models used:

• Tool: OpenAI Codex
• Requested: gpt-5.5
• Model: GPT-5.5 (gpt-5.5)

📎 Log file uploaded as Repository (59333KB)

• View complete solution draft log

---

Now working session is ended, feel free to review and add any feedback on the solution draft.

[konard]

✅ Ready to merge

This pull request is now ready to be merged:

• All CI checks have passed
• No merge conflicts
• No pending changes

---

Monitored by hive-mind with --auto-restart-until-mergeable flag