feat: support multiple PDF uploads per generation

[YizukiAme]

Summary

Currently only a single PDF can be uploaded per generation. Users want to upload multiple PDFs and have them merged as course material.

Current Behavior

• pdfFile: File | null — single file state
• One pdfStorageKey in IndexedDB
• One OCR parsing pass in generation-preview
• One pdfText fed to the generation pipeline

Desired Behavior

• Upload multiple PDF documents
• Merged content is used as course material for generation
• UI shows a file list with individual remove buttons

Design Decisions Needed

• Content merging strategy: concatenate in upload order? per-file section headers?
• Token budget: how to distribute MAX_PDF_CONTENT_CHARS across files — equal split, proportional to length, or sequential truncation?
• OCR images: per-file limit or shared MAX_VISION_IMAGES pool?
• Max file count: cap at N files to avoid overwhelming the LLM context?
• UI: drag-to-reorder? total size indicator?

Affected Layers

Layer	Change
Toolbar UI	File | null → File[], list display
IndexedDB storage	Multiple storage keys or array storage
Session state	Multi-file metadata
Generation preview	Loop OCR parsing + content merge
Generation pipeline	Combined text with truncation

[suzaku007]

may i know when this feature will go live?

[YizukiAme]

Hi @suzaku007, thanks for your interest!

I've already implemented this feature in #160, but the team hasn't scheduled a merge yet, so I can't give you a definite timeline.

If you need it urgently, feel free to cherry-pick the commits from that PR into your own fork. Hope that helps!