fix: handle PG INTERVAL type, robust compact LLM output + feedback

INTERVAL handling
- pg_value_to_json now decodes PG INTERVAL via PgInterval and renders
  it psql-style: `1 year 2 mons 3 days 04:05:06`. Previously
  AVG(timestamp - timestamp) and similar interval-returning queries
  showed `<unsupported type: INTERVAL>` in chat results.
- 7 unit tests covering zero, days-only, mixed, negative, microsecond
  fraction, and the singular/plural unit rules.

Compact reliability
- Sharper system prompt: explicitly instructs plain text starting with
  `-`, no JSON, no fences, no field names. qwen3-coder is heavily
  trained on the agent JSON protocol and was sometimes returning
  `{"action":"final","text":"..."}` even for the compact prompt.
- New clean_summary helper strips ``` fences (with or without lang
  identifier) and extracts the underlying string from a JSON envelope
  if the model still wraps the answer (looks for text/summary/content/
  answer/output keys). 6 unit tests.
- Frontend useChat.compact: success/no-op/error toasts via sonner so
  the user sees what happened. "Nothing to compact" appears when there
  is no older history beyond the last user turn (previously silent).

Verification: cargo test --lib 66 pass (+13), tsc clean, vitest 20
pass.

src-tauri/src/commands/chat.rs

@@ -748,6 +748,40 @@ fn render_thread_for_summary(messages: &[ChatMessage]) -> String {
     out
 }
 
+/// Strip JSON envelopes, markdown fences, and known field-extraction patterns
+/// that the agent-trained model tends to emit even for non-agent prompts.
+/// Returns the underlying summary text.
+fn clean_summary(raw: &str) -> String {
+    let trimmed = raw.trim();
+    if trimmed.is_empty() {
+        return trimmed.to_string();
+    }
+
+    // Strip ```...``` fences (with or without lang).
+    let unfenced = if trimmed.starts_with("```") {
+        let body = trimmed.trim_start_matches('`').trim_start_matches('`').trim_start_matches('`');
+        // Drop optional language identifier on the first line.
+        let after_lang = body.split_once('\n').map(|(_, rest)| rest).unwrap_or(body);
+        let trimmed_end = after_lang.trim_end_matches('`').trim_end_matches('`').trim_end_matches('`');
+        trimmed_end.trim().to_string()
+    } else {
+        trimmed.to_string()
+    };
+
+    // If the model returned a JSON envelope, extract a known string field.
+    if unfenced.starts_with('{') {
+        if let Ok(v) = serde_json::from_str::<serde_json::Value>(&unfenced) {
+            for key in ["text", "summary", "content", "answer", "output"] {
+                if let Some(s) = v.get(key).and_then(|x| x.as_str()) {
+                    return s.trim().to_string();
+                }
+            }
+        }
+    }
+
+    unfenced
+}
+
 /// Find the index of the last User message; returns messages.len() if no user message.
 fn last_user_turn_index(messages: &[ChatMessage]) -> usize {
     for (i, m) in messages.iter().enumerate().rev() {
@@ -793,8 +827,12 @@ pub async fn chat_compact(
          Produce a SHORT summary in the SAME language the user spoke. \
          Use 3-6 bullet points covering: the user's goal, key tables/columns/queries used, \
          numerical findings, conclusions reached, any open questions. \
-         Be concrete with numbers and identifiers. Total length < 800 chars. \
-         Output the bullets directly with no preamble, no JSON, no markdown fences.";
+         Be concrete with numbers and identifiers. Total length < 800 chars.\n\
+         \n\
+         OUTPUT FORMAT: PLAIN TEXT. Start each bullet with `- `. \
+         DO NOT output JSON. DO NOT wrap output in `{` or `}`. \
+         DO NOT use markdown fences. DO NOT include field names like `action`, `text`, `summary`. \
+         DO NOT add a preamble. The first character of your reply must be `-`.";
 
     let llm_messages = vec![
         OllamaChatMessage {
@@ -810,7 +848,7 @@ pub async fn chat_compact(
         .await
         .map_err(|e| TuskError::Ai(format!("Compact failed: {}", e)))?;
 
-    let cleaned = summary.trim();
+    let cleaned = clean_summary(&summary);
     let compacted_msg = ChatMessage::Assistant {
         id: new_id("asst"),
         text: format!(
@@ -1033,6 +1071,42 @@ mod tests {
         assert_eq!(last_user_turn_index(&msgs), msgs.len());
     }
 
+    #[test]
+    fn clean_summary_passes_plain_text() {
+        let s = "- bullet one\n- bullet two";
+        assert_eq!(clean_summary(s), s);
+    }
+
+    #[test]
+    fn clean_summary_strips_markdown_fences() {
+        let s = "```\n- bullet\n```";
+        assert_eq!(clean_summary(s), "- bullet");
+    }
+
+    #[test]
+    fn clean_summary_strips_lang_fence() {
+        let s = "```text\n- bullet one\n- bullet two\n```";
+        assert_eq!(clean_summary(s), "- bullet one\n- bullet two");
+    }
+
+    #[test]
+    fn clean_summary_extracts_text_field_from_json_envelope() {
+        let s = r#"{"action":"final","text":"- bullet one\n- bullet two"}"#;
+        assert_eq!(clean_summary(s), "- bullet one\n- bullet two");
+    }
+
+    #[test]
+    fn clean_summary_extracts_summary_field() {
+        let s = r#"{"summary":"- a\n- b"}"#;
+        assert_eq!(clean_summary(s), "- a\n- b");
+    }
+
+    #[test]
+    fn clean_summary_returns_unchanged_for_unrecognised_json() {
+        let s = r#"{"weird":42}"#;
+        assert_eq!(clean_summary(s), s);
+    }
+
     #[test]
     fn render_thread_for_summary_includes_roles_and_skips_rows() {
         let msgs = vec![