Enhance browser tool with modal dialog handling

## What changed OpenClaw added modal dialog support to its browser tool. Snapshots now display any pending dialogs, and the tool returns a blockedByDialog status when one appears. A new --dialog-id flag lets users target and act on specific dialogs directly from the command line.

The update rolled out on 18 May 2026 as part of the standard CLI release. No extra configuration or paid tier is required.

## Why it matters Modal dialogs break most browser agents on real sites such as login flows, cookie banners, and confirmation prompts. Handling them inside the existing snapshot and command structure removes a common failure point without adding new services.

This move strengthens OpenClaw against closed agents that still stall on pop-ups. It bets that reliable browser control will matter more than raw model size for daily automation work.

## How to use it Pull the latest OpenClaw binary or rebuild from the GitHub repo. Run your normal browser command and inspect the returned snapshot for any blockedByDialog flag.

When a dialog appears, append --dialog-id followed by the ID shown in the snapshot to accept, dismiss, or fill fields. Test first on a non-critical site.

## Watch for Success shows up as fewer manual interventions on sites that previously required human clicks. The bet breaks if dialog detection misses nested or JavaScript-triggered modals. Expect the next step to be file-upload and CAPTCHA handling in the same command surface.