ui: rename Long-form example to Research Report; hyphenate memory labels

- "Long-form research report" example -> "Research Report"
- Memory radio labels and help text use hyphens (discard-all, hide-tool-result) for readability; internal values stay underscored to match MEMORY_STRATEGIES

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

app.py

@@ -1823,7 +1823,7 @@ EXAMPLES = [
         "text": "Find a 2-day itinerary in Tokyo under $250 focused on contemporary art museums and vegetarian restaurants, including transit between sites.",
     },
     {
-        "category": "Long-form research report",
+        "category": "Research Report",
         "icon": "📚",
         "text": "Compare the LLM-safety research approaches of Anthropic, OpenAI, and Google DeepMind over the past 18 months, focusing on alignment techniques and red-teaming methodologies.",
     },
@@ -1916,8 +1916,8 @@ with gr.Blocks(
                     choices=[
                         ("condenser (default)", "condenser"),
                         ("vanilla", "vanilla"),
-                        ("discard_all", "discard_all"),
-                        ("hide_tool_result", "hide_tool_result"),
+                        ("discard-all", "discard_all"),
+                        ("hide-tool-result", "hide_tool_result"),
                     ],
                     value="condenser",
                     elem_id="quest-memory-strategy",
@@ -1926,8 +1926,8 @@ with gr.Blocks(
                     '<div class="memory-help">'
                     '<b>condenser</b> (default) — when context grows large, a State Summarizer LLM compresses earlier turns into a structured JSON of trusted/untrusted/uncertain claims, visited sources, and prior search queries; the agent continues with that compact state.<br>'
                     '<b>vanilla</b> — memory management disabled; the full conversation history is kept.<br>'
-                    '<b>discard_all</b> — when context grows large, the entire message history is reset, restarting the agent from the original question with no accumulated context.<br>'
-                    '<b>hide_tool_result</b> — when context grows large, older tool responses are pruned; only the most recent tool result is kept.'
+                    '<b>discard-all</b> — when context grows large, the entire message history is reset, restarting the agent from the original question with no accumulated context.<br>'
+                    '<b>hide-tool-result</b> — when context grows large, older tool responses are pruned; only the most recent tool result is kept.'
                     '</div>'
                 )
                 max_turns = gr.Slider(