evals: make agent mounting part of tool surfaces

[miguelg719]

Summary

• keep the existing CoreTool abstraction as the canonical tool-surface definition
• attach optional agent delivery and final-state capture to ToolStartResult, preserving one runtime lifecycle
• replace LLMExposure with a lifecycle-free AgentMount value using the independent via: handles | mcp | cli delivery axis
• keep native CoreTool.surface independent from agent delivery, so a code surface may be wrapped by MCP or CLI
• leave task, start URL, and default console binding with the harness
• define reserved harness bindings, CLI environment merge semantics, and capture-before-cleanup ordering

This is stacked on #2590. The revision deliberately defers broad naming/registry churn so the contract change remains a focused two-file diff.

Verification

• pnpm --filter @browserbasehq/stagehand-evals typecheck
• pnpm --filter @browserbasehq/stagehand-evals test:unit (52 files, 413 tests)
• focused formatting check passed
• reviewed independently with Claude Opus 5 against the inline repository diff

[changeset-bot]

⚠️ No Changeset found

Latest commit: d5b3421

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

This PR includes no changesets

When changesets are added to this PR, you'll see the packages that this PR includes changesets for and the associated semver types

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[cubic-dev-ai]

All reported issues were addressed across 21 files

<sub>Reply with feedback, questions, or to request a fix.

Re-trigger cubic</sub>

[cubic-dev-ai]

All reported issues were addressed across 2 files (changes from recent commits).

<sub>Reply with feedback, questions, or to request a fix.

Re-trigger cubic</sub>