chore: restructure llm plugin to be more coherent & fix various bugs in llm ui

Modules/Notifications/Sidebar/Chat/Content/MessageDelegate.qml

@@ -19,7 +19,16 @@ MouseArea {
 	required property ChatListView chatList
 
 	readonly property bool isUser: modelData.role === ChatMessage.Role.User
-	readonly property var blocks: blockify(current.segments, current.error)
+
+	readonly property var blocks: root.isUser ? [
+		{
+			kind: "content",
+			id: 0,
+			text: root.modelData.content,
+			markdown: [],
+			running: false
+		}
+	] : blockify(current.segments, current.error)
 
 	property ChatGeneration current: modelData.activeGeneration
 	property bool reasoningExpanded: false
@@ -54,7 +63,9 @@ MouseArea {
 				blocks.push({
 					kind: "content",
 					id: i,
-					segments: [seg]
+					text: seg.text,
+					markdown: seg.markdown,
+					running: seg.running
 				});
 			} else {
 				const last = blocks.length ? blocks[blocks.length - 1] : null;
@@ -330,11 +341,12 @@ MouseArea {
 						required property var modelData
 
 						isUser: root.isUser
-						segment: modelData.segments[0]
+						text: modelData.text
+						blocks: modelData.markdown
+						running: modelData.running
 						width: root.width
 						repeater: segmentRep
 						message: root.modelData
-						current: root.current
 						hovered: root.containsMouse
 
 						onEdit: text => {