Make most python tests pass with Python 3.4

Dictionary order is not stable in Python < 3.6 so we need to sort
by key to have consistent results.
The LogHandler output is also different on older Python versions.
Also, don't stop running python tests after the first error.

plugins/python/regress/iohelpers.c

@@ -82,7 +82,7 @@ freadall(const char *file_path, char *output, size_t max_len)
     }
 
     if (!feof(file)) {
-        printf("File '%s' was bigger than allocated buffer %lu", file_path, max_len);
+        printf("File '%s' was bigger than allocated buffer %zu", file_path, max_len);
         goto cleanup;
     }