feat: add browser_open and browser_get_text tools

[Rustam-Z]

Summary

Add two new tools to the pyclaudir toolset:

• browser_open — open a URL in a headless browser session
• browser_get_text — extract visible text content from the current browser page

Motivation

These tools would enable the agent to interact with web pages that require JavaScript rendering, login flows, or dynamic content — beyond what WebFetch (static HTTP) can reach.

Proposed behavior

browser_open(url: str) → status

• Opens the given URL in a headless Chromium instance (reuse existing render infra if available)
• Returns load status (success / error / timeout)
• Respects existing private-IP block rules (no localhost, RFC1918, etc.)

browser_get_text(selector?: str) → str

• Returns visible text from the current page
• Optional CSS selector to scope extraction to a specific element
• Truncates at a reasonable limit (e.g. 10k chars) to avoid context overflow

Notes

• Should respect the same internal-URL block rules as WebFetch/WebSearch
• Consider whether to expose as a stateful session (open → interact → close) or stateless (open+get in one call)
• Headless Chromium is already used for render_html — reuse that infra where possible