tests: increased tolerance in fast addition tests

In order to make the test stable increased a tolerance on the size of
data on a node that joins the cluster. The slight increase doesn't
influence the test correctness but makes it less vulnerable to small
changes in size of partition metadata.

Fixes: #20224

Signed-off-by: Michał Maślanka <michal@redpanda.com>

tests/rptest/tests/scaling_up_test.py

@@ -482,7 +482,7 @@ def verify_disk_usage(usage: dict, added_ids: list, percentage: float):
         print_disk_usage(usage)
 
         verify_disk_usage(usage,
-                          [self.redpanda.node_id(self.redpanda.nodes[4])], 0.2)
+                          [self.redpanda.node_id(self.redpanda.nodes[4])], 0.3)
 
         # add sixth node
         self.redpanda.start_node(self.redpanda.nodes[5])
@@ -494,7 +494,7 @@ def verify_disk_usage(usage: dict, added_ids: list, percentage: float):
         verify_disk_usage(usage, [
             self.redpanda.node_id(self.redpanda.nodes[4]),
             self.redpanda.node_id(self.redpanda.nodes[5])
-        ], 0.2)
+        ], 0.3)
         # verify that data can be read
         self.consumer = KgoVerifierSeqConsumer(self.test_context,
                                                self.redpanda,