feat: preserve file extensions in upload URLs

[anoncam]

Summary

Fixes #90 - Adds proper file extension preservation for file uploads and downloads.

When uploading files using dedpaste --file somefile.pdf, the original filename and extension are now preserved in the retrieval URL, allowing proper file type identification upon download.

Changes Made

• CLI Updates: Modified cli/index.ts to send X-Filename header with the basename of uploaded files
• Three upload paths updated:
  1. Send command (both encrypted and non-encrypted)
  2. Default command (encrypted uploads)
  3. Default command (non-encrypted uploads)
• Version bump: Updated to 1.18.0 (minor feature addition)

Technical Details

The server-side code already supported this feature:

• Accepts X-Filename header and stores it in R2 metadata
• Supports optional filename in URL paths: /{id}/{filename} and /e/{id}/{filename}
• Returns URLs with filename when metadata contains it

The gap was that the CLI never sent the filename header. This PR closes that gap.

Before/After Examples

Before

$ dedpaste --file document.pdf
https://paste.d3d.dev/AbCdEfGh

After

$ dedpaste --file document.pdf
https://paste.d3d.dev/AbCdEfGh/document.pdf

$ dedpaste --file photo.jpg --encrypt
https://paste.d3d.dev/e/XyZ123/photo.jpg

$ dedpaste --file data.json --temp
https://paste.d3d.dev/AbCdEfGh/data.json

Use Cases Addressed

• ✅ Binary files (PDFs, images, videos) get proper extensions
• ✅ Automation scripts can determine file handling based on extension
• ✅ Cross-platform sharing with proper file type associations
• ✅ Web browsers can set proper MIME types and Content-Disposition headers

Test Plan

• Build succeeds with TypeScript compilation
• CLI code review confirms proper header injection
• All three upload code paths include the filename header
• Manual testing with actual file uploads (requires deployment)

🤖 Generated with Claude Code