[devchaudhary24k]

Summary

Phases 2 and 3 of implementation.md: the 20 remaining catalogue nodes, their templates and their Studio forms.

• Batch 2: Deduplicate, Auto-trim, Pad / Extend, Pixel-art upscale, Split / Tile, Adjust, Filters, Sharpen / Blur, Background, Text watermark, Image watermark, Border / Rounded corners.
• Batch 3: Set copyright / author, Best format, Responsive set, Favicon / App icons, Placeholder, Palette, Contact sheet, Images to PDF.
• Templates: Watermark and compress, Instagram carousel, Responsive image set.
• Engine: combining nodes can pass their inputs on by reference, so Deduplicate never holds the batch and kept items keep their step cache and source file (ADR 0011). Output and Rename take data and document items; ICO files are documents (ADR 0012). The estimate follows items through Deduplicate.
• Studio: settings forms, card summaries and icons for every new node; the inspector shows the text of data items (palette, snippets, manifest) and tables for Deduplicate and Best format.
• Saving: a save that does not fit in browser storage (pipelines can now hold a logo) shows an error instead of failing silently.
• Tests: the node pair matrix covers all 33 nodes and runs as four shards in parallel.

No new dependencies: BlurHash, the PDF writer and the ICO writer are small in-repo modules.

Testing

• pnpm validate: 1,873 tests and the production build pass. Coverage is 63.5% of statements.
• Every new node has output tests that decode what Output delivered (pixels, PDF structure, ICO entries, BlurHash decoded back, JSON and HTML).
• Engine tests with fake nodes cover passing inputs on, worker pool tests cover the same through OPFS, and preview tests cover data and document items.
• pnpm test:scale (500 images of 12 megapixels) passes.
• On the production build, real photos ran through every new node in the Studio, and every delivered ZIP was checked, including picking an SVG logo and running with it.
• Firefox and WebKit run in CI only.