Correctly handle null check of entry_data_list

I suspect at some point, we were passing in `entry_data_list` as a
`MMDB_entry_data_list_s *`, which is what we do in other extensions, but
that was refactored to `MMDB_entry_data_list_s **` for reasons I do not
recall, and the loop check was never updated appropriately.

This never caused an actual issue as we also have the size check.
However, it could potentially lead to a segfault on a corrupt or
otherwise invalid database.

Author: oschwald

HISTORY.rst

@@ -11,6 +11,8 @@ History
 * The vendored ``libmaxminddb`` has been updated to 1.12.2.
 * The C extension now checks that the database metadata lookup was
   successful.
+* A theoretical segmentation fault with the C extension when doing lookups
+  on a corrupt or invalid database was fixed.
 
 2.6.3 (2025-01-09)
 ++++++++++++++++++

extension/maxminddb.c

@@ -751,10 +751,7 @@ static PyObject *from_map(MMDB_entry_data_list_s **entry_data_list) {
     const uint32_t map_size = (*entry_data_list)->entry_data.data_size;
 
     uint32_t i;
-    // entry_data_list cannot start out NULL (see from_entry_data_list). We
-    // check it in the loop because it may become NULL.
-    // coverity[check_after_deref]
-    for (i = 0; i < map_size && entry_data_list; i++) {
+    for (i = 0; i < map_size && *entry_data_list; i++) {
         *entry_data_list = (*entry_data_list)->next;
 
         PyObject *key = PyUnicode_FromStringAndSize(
@@ -792,10 +789,7 @@ static PyObject *from_array(MMDB_entry_data_list_s **entry_data_list) {
     }
 
     uint32_t i;
-    // entry_data_list cannot start out NULL (see from_entry_data_list). We
-    // check it in the loop because it may become NULL.
-    // coverity[check_after_deref]
-    for (i = 0; i < size && entry_data_list; i++) {
+    for (i = 0; i < size && *entry_data_list; i++) {
         *entry_data_list = (*entry_data_list)->next;
         PyObject *value = from_entry_data_list(entry_data_list);
         if (value == NULL) {