fix(groups): correct get_participants pagination

Two independent defects made every page beyond the first wrong:

1. TelegramClient.iter_participants takes no `offset` kwarg — its signature
   is (entity, limit=None, *, search, filter, aggressive) — so passing
   offset= raised TypeError for page > 1.

2. `limit` does not bound basic groups. _ParticipantsIter._init buffers the
   whole participant list and returns True, and RequestIter.__anext__ then
   sets `left` to the buffer length, discarding the requested limit. A page
   of a basic group therefore ran past page_size.

Skip to the start of the requested page and stop once it is full. `limit`
is still passed so channels, where it is honoured, do not over-fetch.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

Author: 2 people

telegram_mcp/tools/groups.py

@@ -243,12 +243,13 @@ async def get_participants(
         cl = get_client(account)
         await ensure_connected(cl)
 
-        # Use iter_participants with offset to fetch only the needed slice,
-        # avoiding O(N) fetching on later pages.
+        # iter_participants takes no `offset`, and its `limit` is not honoured
+        # for basic groups. Fetch through the page, then slice it out.
         offset = (page - 1) * page_size
         participants = []
-        async for participant in cl.iter_participants(chat_id, limit=page_size, offset=offset):
+        async for participant in cl.iter_participants(chat_id, limit=offset + page_size):
             participants.append(participant)
+        participants = participants[offset : offset + page_size]
 
         if not participants:
             return format_tool_result([])