Filter hidden elements in text extraction to match browser visibility behavior

Problem

Currently, our text extraction from Wikipedia and other websites includes hidden elements that browsers don't render. This causes issues with proof line matching between extracted text and what users see in the GUI.

Example issue: Wikipedia infoboxes contain hidden spans like <span style="display:none">(1977-08-21)</span> that appear in our extracted text but are invisible in browsers.

Current Behavior

Our enrichment pipeline extracts text using:

soup = BeautifulSoup(html_content, "html.parser")
text = soup.get_text()

This includes ALL text content, including hidden elements.

Expected Behavior

Text extraction should match what browsers display, excluding:

• Elements with display: none
• Elements with visibility: hidden
• Elements with hidden attribute
• Script/style/noscript tags (already excluded by GUI highlighter)

GUI Highlighting Logic Reference

Our text highlighter at /poliloom-gui/src/lib/textHighlighter.ts:34-41 already excludes SCRIPT, STYLE, and NOSCRIPT elements but doesn't handle CSS-hidden elements.

Start by locating the enrichment pipeline code containing the shown BeautifulSoup text extraction, then compare it with the exclusions in /poliloom-gui/src/lib/textHighlighter.ts:34-41. The change is done when extracted text excludes CSS-hidden elements, hidden attributes, and script/style/noscript content so proof-line matching follows browser-visible text.