[CI][Bugfix] Add tp_shard_with_padding to padded MoE reload test mock (#57402)

Signed-off-by: kimi-no-na-wa <kimi-no-na-wa@users.noreply.github.com>
Co-authored-by: kimi-no-na-wa <kimi-no-na-wa@users.noreply.github.com>
Co-authored-by: Misha Goin <mgoin64@gmail.com>

tests/model_executor/model_loader/test_reload.py

@@ -489,6 +489,7 @@ def test_padded_moe_reload_releases_each_layer(
         is_act_and_mul=is_gated,
         has_bias=has_bias,
         tp_rank=tp_rank,
+        tp_shard_with_padding=False,
         moe_parallel_config=SimpleNamespace(tp_size=2),
     )
     model = torch.nn.ModuleList()