fix(explain): address review findings — N+1 queries, duplicate decisions, silent errors

1. fetch_open_threads: replace N+1 loop (2 queries per thread) with a
   single query using correlated subqueries for note_count and started_by.
2. extract_key_decisions: track consumed notes so the same note is not
   matched to multiple events, preventing duplicate decision entries.
3. build_timeline_excerpt_from_pipeline: log tracing::warn on seed/collect
   failures instead of silently returning empty timeline.

src/search/fts.rs

@@ -119,15 +119,12 @@ pub fn search_fts(
 }
 
 pub fn generate_fallback_snippet(content_text: &str, max_chars: usize) -> String {
-    if content_text.chars().count() <= max_chars {
-        return content_text.to_string();
-    }
-
-    let byte_end = content_text
-        .char_indices()
-        .nth(max_chars)
-        .map(|(i, _)| i)
-        .unwrap_or(content_text.len());
+    // Use char_indices to find the boundary at max_chars in a single pass,
+    // short-circuiting early for large strings instead of counting all chars.
+    let byte_end = match content_text.char_indices().nth(max_chars) {
+        Some((i, _)) => i,
+        None => return content_text.to_string(), // content fits within max_chars
+    };
     let truncated = &content_text[..byte_end];
 
     if let Some(last_space) = truncated.rfind(' ') {