[BUG] 进行mmengine适配后出现异步bug

[ShaohonChen]

🐛 Bug description [Please make everyone to understand it]

#497

增加了mm框架适配，使用如下代码在mmengine中调用swanlab并开启日志云端同步时出现bug

#  use swanlab
custom_imports = dict(imports=["swanlab.integration.mmengine"], allow_failed_imports=False)
vis_backends = [
    dict(
        type="SwanlabVisBackend",
        save_dir="./swanlab",
        init_kwargs={
            "experiment_name": "use_swanlab_in_mmengine",
            "description": "have fun",
        },
    ),
]
visualizer = dict(
    type="Visualizer",
    # name="MustHaveNameIfJustVis",
    vis_backends=vis_backends,
)

bug日志如下（似乎是云端创建环境时出问题）。

error.txt

🧑‍💻 Step to reproduce

#确保切换到分支线中feat/integration-mmengine中的代码并且构建并安装新的swanlab包。

安装和配置mmdetection的环境，参考mmdetection get start

将如下内容添加到mmdetection中任意config

custom_imports = dict(imports=["swanlab.integration.mmengine"], allow_failed_imports=False)
vis_backends = [
    dict(
        type="SwanlabVisBackend",
        save_dir="./swanlab",
        init_kwargs={
            "experiment_name": "use_swanlab_in_mmengine",
            "description": "have fun",
        },
    ),
]
visualizer = dict(
    type="Visualizer",
    # name="MustHaveNameIfJustVis",
    vis_backends=vis_backends,
)

使用如下命令开启训练

bash tools/dist_train.sh {CONFIG_FILE_PATH} 8

👾 Expected result

如将config文件修改为本地运行模式即可正常工作.

custom_imports = dict(imports=["swanlab.integration.mmengine"], allow_failed_imports=False)
vis_backends = [
    dict(
        type="SwanlabVisBackend",
        save_dir="./swanlab",
        init_kwargs={
            "experiment_name": "use_swanlab_in_mmengine",
            "description": "have fun",
            "cloud": False,
        },
    ),
]
visualizer = dict(
    type="Visualizer",
    # name="MustHaveNameIfJustVis",
    vis_backends=vis_backends,
)

🚑 Any additional [like screenshots]

• SwanLab Version:

• Platform:

[SAKURA-CAT]

在 #515 中从异步换成了同步

[SAKURA-CAT]

容我测试完毕后push到main

[SAKURA-CAT]

fix by #515