Support local source for sagemaker.core.shapes.ProcessingInput

**Describe the feature you'd like**
Allow for `sagemaker.core.shapes.ProcessingInput` to accept local file path as input, for automatic uploading of input files to S3.

**How would this feature be used? Please describe.**
When migrating from sagemaker v2 to v3, the `sagemaker.processing.ProcessingInput` used in `sagemaker.workflow.steps.ProcessingStep.inputs()` is no longer applicable, replaced by `sagemaker.core.shapes.ProcessingInput` used in `sagemaker.core.processing.ScriptProcessor.run()`. `sagemaker.core.shapes.ProcessingInput` accepts only S3 path, which makes it more convoluted for the training pipeline process as the files involved in the pipeline have to be uploaded separately before the creation of the pipeline, instead of how it was done in v2's `sagemaker.processing.ProcessingInput` which uploads it while the pipeline is being built.

首先跟踪 sagemaker.core.shapes.ProcessingInput 和 sagemaker.core.processing.ScriptProcessor.run()，然后将它们的行为与 issue 中描述的 v2 sagemaker.processing.ProcessingInput 进行比较。确定在构建 pipeline 时应如何上传本地文件路径，并验证现有的 S3 输入仍能正常工作。

由索引模型根据 Issue 内容生成。