Site Editor Benchmark: add custom WordPress environment support

[gcsecsey]

Related issues

Related to STU-1305
Follow up to #2600

Proposed Changes

Adds the ability to benchmark any running WordPress site, not just Studio and Playground environments. This PR:

• Adds a repeatable --custom flag. A running WordPress site can be added to the benchmark run via --custom=<name>,<url>[,<user>,<password>]. The flag is repeatable, so multiple custom sites can be compared in a single run (eg. bare vs with plugins). Username defaults to admin, password to password.
• Adds an --install-plugins flag. This installs the plugins in the plugins-blueprint.json file using the standard WordPress REST API (cookie auth + nonce). --install-plugins installs plugins for all custom environments, or --install-plugins=<name1>,<name2> targets specific ones.
• Updates the navigation to handle setup wizard redirects. Custom environments log in via the WordPress login form, then navigate to /wp-admin/ explicitly.

Examples

# Benchmark a single custom site
npm run benchmark -- --custom=my-site,http://localhost:10003

# Compare bare vs. with plugins
npm run benchmark -- \
  --custom=custom-bare,http://localhost:10003 \
  --custom=custom-plugins,http://localhost:10004 \
  --install-plugins=custom-plugins

# Compare Studio vs. a custom site
npm run benchmark -- --only=studio,my-site \
  --custom=my-site,http://localhost:10003 --rounds=3

Testing Instructions

• Set up a WordPress site (Local, Docker, MAMP, etc.) with an admin account
• Run a bare custom benchmark:

  cd tools/benchmark-site-editor
  npm install
  npm run benchmark -- --skip-studio --skip-playground-cli --skip-playground-web \
    --custom=my-site,http://localhost:10003

• Check that the benchmark completes and prints results
• Run with --install-plugins and verify plugins are installed
• Run again to verify already-installed plugins are skipped
• Run with multiple --custom flags to verify multiple sites work
• Verify existing environments still work: npm run benchmark -- --only=studio

Pre-merge Checklist

• TypeScript compilation clean
• Plugin installation works via REST API with proper activation hooks
• Teardown guaranteed even on plugin install failure
• Local environments excluded by default

[wpmobilebot]

📊 Performance Test Results

Comparing e828ddc vs trunk

site-editor

Metric	trunk	e828ddc	Diff	Change
load	1436.00 ms	1417.00 ms	-19.00 ms	⚪ 0.0%

site-startup

Metric	trunk	e828ddc	Diff	Change
siteCreation	7093.00 ms	7086.00 ms	-7.00 ms	⚪ 0.0%
siteStartup	3932.00 ms	3937.00 ms	+5.00 ms	⚪ 0.0%

---

Results are median values from multiple test runs.

Legend: 🟢 Improvement (faster) | 🔴 Regression (slower) | ⚪ No change (<50ms diff)