Reels pre-gen: record true media count + real upsert for user_ai_prefs

- pregen_one recorded media_count as planned.len() (beat count); record
  the actual media item total (media.len(), photos + clips) in both the
  cache-hit and freshly-rendered ledger paths. Drops the redundant
  photo_count binding.
- Replace upsert_prefs's insert-then-catch-error-then-update dance with a
  single atomic INSERT ... ON CONFLICT(id) DO UPDATE. Explicit id=1 makes
  the conflict target deterministic; explicit column .set((...)) keeps
  None -> NULL overwrite semantics so the row mirrors the latest request
  exactly, and genuine insert errors surface instead of being swallowed.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

src/reels/mod.rs

@@ -1075,6 +1075,8 @@ async fn pregen_one(
     // Flatten every media item across beats (in order) into the cache key.
     let media: Vec<SegmentMedia> = planned.iter().flat_map(|b| b.media.clone()).collect();
     let key = cache_key(&selector, &media, voice.as_deref());
+    // Total media items shown (photos + clips), not beat count.
+    let media_count = media.len() as i32;
 
     // Dedup: check if fresh ledger row exists
     let now = std::time::SystemTime::now()
@@ -1127,7 +1129,7 @@ async fn pregen_one(
                 cache_key: key.clone(),
                 output_path: mp4_path.to_string_lossy().to_string(),
                 title,
-                media_count: planned.len() as i32,
+                media_count,
                 render_version: RENDER_VERSION as i32,
                 tz_offset_minutes: tz,
                 voice: voice.clone(),
@@ -1139,7 +1141,6 @@ async fn pregen_one(
 
     // Generate the reel
     log::info!("Generating precomputed reel for span={}, key={}", span, key);
-    let photo_count = planned.len() as i32;
     let (title, mp4) = produce_reel(
         app_state,
         insight_dao,
@@ -1163,7 +1164,7 @@ async fn pregen_one(
             cache_key: key.clone(),
             output_path: mp4.to_string_lossy().to_string(),
             title,
-            media_count: photo_count,
+            media_count,
             render_version: RENDER_VERSION as i32,
             tz_offset_minutes: tz,
             voice: voice.clone(),