Browser Agent v0.4.32

Fix: DOM extraction field name mismatch and fallback_js support

• extract_dom now accepts both name and field as extraction identifiers — fixes silent extraction failures where the template used name but the code only accepted field
• Added fallback_js support — when a CSS selector fails to find a value, the executor now tries the fallback_js JavaScript expression as a fallback
• Improved logging during DOM extraction for easier debugging