增加重型SPA页面的超时逻辑，增加raw mode避免cli页面agent猜测变量名的行为

[YoungCan-Wang]

Description

Related issue:

Type of Change

• 🐛 Bug fix
• ✨ New feature
• 🌐 New site adapter
• 📝 Documentation
• ♻️ Refactor
• 🔧 CI / build / tooling

Checklist

• I ran the checks relevant to this PR
• I updated tests or docs if needed
• I included output or screenshots when useful

Documentation (if adding/modifying an adapter)

• Added doc page under docs/adapters/ (if new adapter)
• Updated docs/adapters/index.md table (if new adapter)
• Updated sidebar in docs/.vitepress/config.mts (if new adapter)
• Updated README.md / README.zh-CN.md when command discoverability changed
• Used positional args for the command's primary subject unless a named flag is clearly better
• Normalized expected adapter failures to CliError subclasses instead of raw Error

Screenshots / Output

[YoungCan-Wang]

Quick maintainer summary for this PR:

• Main functional change: allow structured navigateBefore options (url, waitUntil, settleMs) so heavy SPA adapters can avoid the default pre-nav load/settle behavior when that causes timeouts.
• Also adds --raw to print unformatted JSON for adapter development/debugging.
• Also adds typed errors (NetworkError, ApiError, PageChangedError) and updates the documented exit-code mapping.

Current head is mergeable and the latest checks on 4332ab8 are green:

• Doc Check
• Security Audit
• CI
• E2E Headed Chrome

If you prefer, I can split --raw / error-class additions into a separate PR and keep this PR focused only on the SPA pre-navigation timeout change.

[jackwener]

@YoungCan-Wang Thanks for the work on this — revisiting after some time, the core problems this PR addresses are already handled on main:

1. Heavy SPA pre-nav timeout: page.goto(url, { waitUntil: 'none', settleMs }) is already on main (src/browser/page.ts). Adapters that need custom pre-nav behavior can set navigateBefore: false and call page.goto() with those options directly in func() — no framework change needed. The declarative navigateBefore: { url, waitUntil, settleMs } object form you add here would be sugar on top, not a blocker.
2. Typed errors (NetworkError / ApiError / PageChangedError): Existing CliError('NETWORK', ...) with custom codes already produces the same exit-code behavior. Adding named subclasses is a normalization rather than a fix for a broken path.
3. --raw output: This one isn't on main, but --format json covers most of the adapter-development use case through the columns filter. Happy to take this as a separate focused PR if you'd like to pick it up.

Closing this since there's no blocking issue left. Really appreciate the contribution and sorry for the delay on review — if you want to carve out --raw as its own smaller PR, that's very welcome.