Add clarifying comments about paging behavior

Explain that system.local always has exactly 1 row and will never
require additional pages, while system.peers might need multiple
pages only for very large clusters (>1000 nodes). This addresses
the concern about serial page fetching - in practice, there's no
performance degradation because:
1. First pages are fetched in parallel via wait_for_responses
2. system.local never has additional pages (always 1 row)
3. system.peers rarely has additional pages (only for >1000 nodes)
4. Additional pages within a query must be serial (CQL protocol)

Co-authored-by: mykaul <4655593+mykaul@users.noreply.github.com>

Author: Copilot

cassandra/cluster.py

@@ -3660,6 +3660,8 @@ def _try_connect(self, host):
                     peers_query, timeout=self._timeout)
 
             # Fetch all pages if there are more results
+            # Note: system.local always has exactly 1 row, so it will never have additional pages
+            # system.peers might have multiple pages for very large clusters (>1000 nodes)
             peers_result = self._fetch_all_pages(connection, peers_result, peers_query, self._timeout)
             local_result = self._fetch_all_pages(connection, local_result, local_query, self._timeout)
 
@@ -3841,6 +3843,8 @@ def _refresh_node_list_and_token_map(self, connection, preloaded_results=None,
                 peers_query, local_query, timeout=self._timeout)
 
             # Fetch all pages if there are more results
+            # Note: system.local always has exactly 1 row, so it will never have additional pages
+            # system.peers might have multiple pages for very large clusters (>1000 nodes)
             peers_result = self._fetch_all_pages(connection, peers_result, peers_query, self._timeout)
             local_result = self._fetch_all_pages(connection, local_result, local_query, self._timeout)
 
@@ -4148,6 +4152,8 @@ def wait_for_schema_agreement(self, connection=None, preloaded_results=None, wai
                         peers_query, local_query, timeout=timeout)
 
                     # Fetch all pages if there are more results
+                    # Note: system.local always has exactly 1 row, so it will never have additional pages
+                    # system.peers might have multiple pages for very large clusters (>1000 nodes)
                     peers_result = self._fetch_all_pages(connection, peers_result, peers_query, timeout)
                     local_result = self._fetch_all_pages(connection, local_result, local_query, timeout)
                 except OperationTimedOut as timeout: