[SPARK-15550][SQL] Dataset.show() should show contents nested products as rows

## What changes were proposed in this pull request?

This PR addresses two related issues:

1. `Dataset.showString()` should show case classes/Java beans at all levels as rows, while master code only handles top level ones.

2. `Dataset.showString()` should show full contents produced the underlying query plan

   Dataset is only a view of the underlying query plan. Columns not referred by the encoder are still reachable using methods like `Dataset.col`. So it probably makes more sense to show full contents of the query plan.

## How was this patch tested?

Two new test cases are added in `DatasetSuite` to check `.showString()` output.

Author: Cheng Lian <lian@databricks.com>

Closes #13331 from liancheng/spark-15550-ds-show.