fix(ci): add Bun tarballs and Opus 5 to PR automation

[tejaskash]

Summary

• add authorized PR tarball generation to the consolidated PR automation workflow
• use Bun for dependency installation, build, packaging, and install guidance
• write the tarball to the runner temporary directory and verify the file before release creation
• set Harness reviews to the active us.anthropic.claude-opus-5 profile

Verification

• go run github.com/rhysd/actionlint/cmd/actionlint@v1.7.12 .github/workflows/pr-automation.yml
• bunx prettier --check .github/workflows/pr-automation.yml
• bun pm pack --destination "$RUNNER_TEMP" equivalent with a real-file check
• bun test with 1,932 passing tests
• bun run typecheck
• bun run lint:check
• bun run format:check
• bun run build

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 97.33%. Comparing base (343d213) to head (0c58ed1).

Additional details and impacted files

@@            Coverage Diff            @@
##           refactor    #2100   +/-   ##
=========================================
  Coverage     97.33%   97.33%           
=========================================
  Files           417      417           
  Lines         25250    25250           
=========================================
  Hits          24578    24578           
  Misses          672      672           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[agentcore-devx-automation]

AgentCore Harness Review

Verdict: Looks good

Single-line CI change that sets model_id: us.anthropic.claude-opus-5 on the reusable reusable-pr-ai-review.yml workflow so Harness PR reviews use Opus. The rest of the diff versus main is just YAML reflow of pr_url and quote-style differences that don't affect behavior. The referenced system_prompt_path/review_prompt_path files exist on the PR branch under .github/harness/prompts/, so nothing is dangling.

The only thing worth double-checking is that us.anthropic.claude-opus-5 is the exact inference-profile ID the reusable workflow / underlying Bedrock account expects (other places in this repo use fully-qualified IDs like us.anthropic.claude-sonnet-4-5-20250929-v1:0). If it isn't recognized, PR reviews will just start failing until it's corrected, so the blast radius is small. Nothing blocking.

[agentcore-devx-automation]

Claude Security Review: no high-confidence findings. (run)

[agentcore-devx-automation]

Claude Security Review: no high-confidence findings. (run)

[agentcore-devx-automation]

Claude Security Review: no high-confidence findings. (run)