Upload pipeline: staging, MD5 auto-match, IQDB, three-column board

Backend:
- TempUpload model: staged files (pending / visual_match / completed /
  error) with resolution, e621 payload, custom metadata and IQDB data
- Files land in a temp folder and only move into the watched library
  folder once resolved; duplicates resolve immediately without a copy
- Endpoints: stage (multipart), list, retrieve, temp file, IQDB save,
  resolve (link to post or custom metadata), discard/dismiss
- cleanup_temp_uploads command for old staged files
- Replaces the old direct-to-library upload endpoint

Frontend:
- Upload page is now a three-column board (Pending & Unmatched /
  Visual Similarity Detected / Auto-uploaded & Indexed)
- After upload: MD5s are batch-checked against e621 and matches
  auto-complete with full post metadata; remaining files run through
  IQDB and move to the similarity column when candidates exist
- Metadata modal with IQDB candidates, post-ID linking and custom
  tags/rating/notes; discard and dismiss actions
- e621 client gains fetchPostsByMd5 and iqdbSearch helpers

Roadmap updated with the completed upload items.

ROADMAP.md

@@ -37,32 +37,31 @@ they land.
   - [ ] "Followed" nav entry
 - [ ] `F` shortcut to favorite a post, tag finder in the Ctrl+K palette
 
-## 3. Uploads & ingestion pipeline (major rework)
+## 3. Uploads & ingestion pipeline
 
-Current upload writes straight into the watched folder. The target flow matches
-the original app:
+Files now stage first and are resolved before entering the library.
 
-- [ ] **Backend staging storage**
-  - [ ] `TempUpload` model: temp_id, user, md5, filename, temp path, status
-        (`pending` / `visual_match` / `completed` / `error`), timestamps
-  - [ ] Files land in a temp folder first; only "completed" moves them into the
-        watched library folder
-  - [ ] Pending uploads list endpoint, delete/discard endpoint
-- [ ] **Auto-upload / auto-match**
-  - [ ] Compute MD5 on upload; if it already exists in the library, mark as
-        duplicate and resolve automatically (no new file)
-  - [ ] If the MD5 matches an e621 post, auto-complete: fetch and store the post
-        metadata, seed rating/tags, mark as indexed
-  - [ ] Batch MD5 checks against e621 for staged uploads
-- [ ] **IQDB / visual similarity on upload**
-  - [ ] Run IQDB on staged uploads and surface candidate posts
-  - [ ] Perceptual-hash check against the library → "Visual Similarity Detected"
-        state with side-by-side comparison
-- [ ] **Upload UI**
-  - [ ] Three-column board: Pending & Unmatched / Visual Similarity Detected /
+- [x] **Backend staging storage**
+  - [x] `TempUpload` model: user, md5, filename, temp path, status
+        (`pending` / `visual_match` / `completed` / `error`), resolution
+  - [x] Files land in a temp folder first; only completed uploads move into
+        the watched library folder
+  - [x] Pending/upload list endpoint, file serving, discard endpoint
+  - [x] `cleanup_temp_uploads` command for old staged files
+- [x] **Auto-upload / auto-match**
+  - [x] MD5 computed on staging; exact duplicates resolve immediately
+  - [x] MD5 batch-checked against e621; matches auto-complete with post
+        metadata stored and rating seeded
+- [x] **IQDB similarity on upload** (SPA-driven)
+  - [x] Automatic + manual IQDB checks with candidate posts
+  - [x] "Visual Similarity Detected" state with candidate picker
+  - [ ] Perceptual-hash comparison against the library (shared with the
+        duplicates engine, still pending)
+- [x] **Upload UI**
+  - [x] Three-column board: Pending & Unmatched / Visual Similarity Detected /
         Auto-uploaded & Indexed
-  - [ ] Metadata modal (link to e621 post via search + custom tags/rating/notes)
-  - [ ] Per-file progress plus batch summary
+  - [x] Metadata modal (link to e621 post, IQDB candidates, custom metadata)
+  - [x] Per-file progress plus batch processing indicator
 
 ## 4. Staff tools