feat(chat): replace model picker with situation modes (Chat/Analysis/Graph)

The composer's SELECT is now a clinician-facing mode picker instead of a list
of LLM vendors (the model still comes from Settings → AI and continues to drive
the Veil consent gate). The mode travels with each send:

- backend appends an Analysis/Graph directive to the agent's system prompt
- Graph mode renders an Obsidian-style RecordGraph for the /patient fast-path
  (new data-recordGraph message part reusing the shared graph component)

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

frontend/lib/chat-modes.ts

@@ -0,0 +1,44 @@
+// The chat "situation" modes shown in the composer — what the clinician wants
+// the assistant to do, rather than which LLM runs (the model/provider is set
+// once in Settings → AI). The mode travels with each send so the backend can
+// shape its system prompt, and Graph mode also renders the record graph for the
+// `/patient` fast-path.
+
+import { MessageSquare, Network, Sparkles } from "lucide-react";
+
+export type ChatMode = "chat" | "analysis" | "graph";
+
+export const DEFAULT_MODE: ChatMode = "chat";
+
+export type ChatModeOption = {
+  id: ChatMode;
+  icon: typeof MessageSquare;
+  // i18n keys under `chat.input.modes.*`.
+  labelKey: string;
+  descriptionKey: string;
+};
+
+export const CHAT_MODES: ChatModeOption[] = [
+  {
+    id: "chat",
+    icon: MessageSquare,
+    labelKey: "chat.input.modes.chat.label",
+    descriptionKey: "chat.input.modes.chat.description",
+  },
+  {
+    id: "analysis",
+    icon: Sparkles,
+    labelKey: "chat.input.modes.analysis.label",
+    descriptionKey: "chat.input.modes.analysis.description",
+  },
+  {
+    id: "graph",
+    icon: Network,
+    labelKey: "chat.input.modes.graph.label",
+    descriptionKey: "chat.input.modes.graph.description",
+  },
+];
+
+export function getMode(id: string): ChatModeOption {
+  return CHAT_MODES.find((m) => m.id === id) ?? CHAT_MODES[0];
+}