Changes in v2.0.0:
Added 3 export modes (switchable via tabs):

Grid PNG (original mode) — Merges images into a single grid using Canvas API
ZIP — Packages all images at full resolution with zero quality loss
PDF — Combines images into a multi-page PDF, one image per page at full resolution

New dependencies:

fflate (v0.8.2) — Fast ZIP compression library, bundled locally in sidebar/libs/
jsPDF (v2.5.1) — PDF generation library, bundled locally in sidebar/libs/

Why these modes were added:
The original grid PNG compresses 9 images into 1800×1800px, which works great for photos but makes text blurry in screenshots containing small text. ZIP and PDF modes solve this by preserving full original resolution.
No external network calls:
Both libraries are bundled locally. No CDN requests, no data transmission. All processing happens client-side.
User experience:
ZIP and PDF modes include a copy-to-clipboard prompt that users can paste alongside the file when uploading to AI chats (e.g., "Here are 9 images in a ZIP, please unzip and analyze"). This helps AI tools properly handle the multi-image input.
All functionality remains 100% local — no data collection, no external servers.