feat(backend): feed lists name, username and photo changes of people and chat renames, merged per day

backend/src/api/mcp/server.py

@@ -78,7 +78,8 @@ Typical flows:
   joins and leaves, deleted message count, the private chat.
 - Forensics: `get_feed` lists what was deleted or edited in Telegram across
   all chats, newest first, with the text before and after each edit; a bulk
-  deletion is one item with `message_count`. `get_deleted_messages` and
+  deletion is one item with `message_count`. Its `profile` kind lists name,
+  username and photo changes of people and renames of chats. `get_deleted_messages` and
   `get_message_versions` recover the content of one chat or one message.
 - Files: `download_media` gets any message attachment (photo, video, voice,
   document, ...). It pulls the file from Telegram if the archive lacks it and
@@ -270,14 +271,17 @@ async def get_feed(
     limit: int = DEFAULT_LIMIT,
     cursor: str | None = None,
 ) -> dict[str, Any]:
-    """Deleted and edited messages across chats, newest first.
+    """Deleted and edited messages and profile changes across chats, newest first.
 
-    `kind` picks deleted, edited or all; `since` keeps items at or after that
-    time; `chat_id` or `chat_kind` (dm, group, channel) narrow it down. Edited
-    items carry `before` and `after` texts. A deleted item with
-    `message_count` above 1 is a bulk deletion in one chat; read its messages
-    with `get_deleted_messages` for that chat. Pass `next_cursor` back as
-    `cursor` for the next page.
+    `kind` picks deleted, edited, profile, messages (deleted and edited) or all;
+    `since` keeps items at or after that time; `chat_id` or `chat_kind` (dm,
+    group, channel) narrow it down. Edited items carry `before` and `after`
+    texts. A deleted item with `message_count` above 1 is a bulk deletion in
+    one chat; read its messages with `get_deleted_messages` for that chat.
+    Profile items cover people in private chats or contacts and chats in the
+    dialog list: `detail` maps `name`, `username`, `photo` (unique file ids) or
+    `title` to `[old, new]`, changes a day apart or closer are merged. Pass
+    `next_cursor` back as `cursor` for the next page.
     """
     if cursor is not None and re.fullmatch(CURSOR_PATTERN, cursor) is None:
         msg = "cursor must be a next_cursor returned by get_feed"