Support Conditional Hyperparameter in AMT

**Describe the feature you'd like**

Hyperparameter tuning frameworks like [optuna](https://optuna.org/) and google's [vertex AI](https://cloud.google.com/vertex-ai/docs/training/hyperparameter-tuning-overview#conditional_hyperparameters) all support some form of conditional hyperparameter. Could you consider supporting this feature in some capacity?

**How would this feature be used? Please describe.**

For instance, in training a cnn, we can define hyperparameters conditional on their parent hyperparameter matching some condition:

```
search_space = {}
if optimizer == 'adam':
search_space['adam_beta_1'] = range(...)
search_space['adam_beta_2'] = range(...)
elif optimizer == 'sgd':
...
```

**Describe alternatives you've considered**

Right now there are a few ways to work around this:

* Using optuna with sagemaker
* Enumerate all possible hyperparameters and handle appropriately in the training entry point (i.e., ignore the ones not needed)

It would be very helpful if this feature is supported by default. Thanks.

未指定任何文件、测试或入口点。首先查看现有的 AMT 超参数调优流程以及链接的 Optuna 和 Vertex AI 条件超参数文档，然后确定 CNN 示例的条件搜索空间应如何表示和验证。

由索引模型根据 Issue 内容生成。