`ProcessingS3Output`'s `s3_uri` to be an optional field

**Describe the feature you'd like**
In the SDK v2, `ProcessingOutput` has a `destination` parameter to be either an S3 path where users want to write their results, but they can set it to `None` when they want to delegate the storage to SageMaker. This option stores the results in an S3 prefix partitioned by `job_name`, `step_name`, and `output_name`, which is quite convenient for traceability and to store intermediate results of a pipeline.

**How would this feature be used? Please describe.**
Much like v2, v3 should also support this feature. Users may not want to fix the destination path for intermediate results, so ideally the same behaviour should apply to v3 by leaving `ProcessingS3Output`'s `s3_uri` field as `None`. Currently this field is mandatory and it's unclear whether users can reproduce the v2 `None` behaviour.

**Describe alternatives you've considered**
No alternatives considered so far other than sticking to v2.

**Additional context**
N/A

首先比较 SDK v2 ProcessingOutput 的行为与 v3 ProcessingS3Output 的定义及其对目标值的处理。确认对于中间输出，应如何表示省略的 S3 URI；完成条件是 v3 支持文档中所述的 v2 等效用例，并通过适当的测试验证该行为。

由索引模型根据 Issue 内容生成。