Phase 4: Ask Petal SSE chat

Conversational follow-up on a suggestion, streamed token-by-token.

Backend (interface-only; handlers never touch a concrete LLM client):
- internal/llm/chat.go: StreamAskPetal with conversational sampling
  (max_tokens 512, temp 0.7, rep 1.15, top_p 0.92, stop "\n\n\n"),
  reusing AskPetalSystemPrompt + TrimHistory.
- internal/suggestions/chat.go: POST /api/suggestions/:id/chat. One
  user-scoped join loads the suggestion + parent content_text;
  surroundingParagraph extracts the \n\n-bounded paragraph at from_pos
  (whole-doc fallback when unlocated) and injects it server-side.
  Streams event: token / event: done SSE frames with JSON-encoded data
  so token newlines can't break framing; real http.Flusher per chunk.
  LLM-unreachable -> 502 before SSE headers; unknown suggestion -> 404.

Frontend:
- streamSuggestionChat: fetch + ReadableStream SSE parser (not
  EventSource, needs POST), abortable.
- AskPetal.tsx: whole conversation in component state (no persistence,
  cleared on close), Petal's first bubble pre-seeded with the
  explanation, rose/lavender bubbles, CJK font stack on the bubbles
  only (Note #17), streaming caret.
- SuggestionCard "Ask Petal" pill pins the card open while chatting
  (hover-close suppressed, click-away closes) and widens it to 340px.

Tests: chat_test.go covers streamed-text concat + done event,
server-side context injection on the system message, sampling params,
404, and surroundingParagraph. go build/vet/test clean, tsc clean,
vite build OK. Live SSE smoke-tested against a fake streaming vLLM:
tokens flushed individually through the chi middleware stack, done
terminator, 502 on LLM-down, 404 on unknown suggestion.

Claude-Session: https://claude.ai/code/session_016Yr6jELuRc7hyzYLccQKZd

internal/llm/chat.go

@@ -0,0 +1,29 @@
+package llm
+
+import "context"
+
+// AskPetalMessages assembles the message array for one Ask Petal turn: the
+// suggestion-context system prompt followed by the (length-capped) client-side
+// conversation history. The backend is stateless, so the full history rides on
+// every request (spec: no server-side sessions).
+func AskPetalMessages(systemPrompt string, history []Message) []Message {
+	msgs := make([]Message, 0, len(history)+1)
+	msgs = append(msgs, Message{Role: "system", Content: systemPrompt})
+	msgs = append(msgs, TrimHistory(history)...)
+	return msgs
+}
+
+// StreamAskPetal opens an SSE-friendly token stream for an Ask Petal reply using
+// the conversational sampling parameters from the spec. The caller forwards the
+// returned chunks to the browser; the channel closes when generation ends.
+func StreamAskPetal(ctx context.Context, client LLMClient, systemPrompt string, history []Message) (<-chan string, error) {
+	return client.Stream(ctx, CompletionRequest{
+		Messages:          AskPetalMessages(systemPrompt, history),
+		MaxTokens:         512,
+		Temperature:       0.7,
+		RepetitionPenalty: 1.15,
+		TopP:              0.92,
+		Stop:              []string{"\n\n\n"},
+		Stream:            true,
+	})
+}