[SPARK-15443][SQL] Fix 'explain' for streaming Dataset

## What changes were proposed in this pull request?

- Fix the `explain` command for streaming Dataset/DataFrame. E.g.,
```
== Parsed Logical Plan ==
'SerializeFromObject [staticinvoke(class org.apache.spark.unsafe.types.UTF8String, StringType, fromString, input[0, java.lang.String, true], true) AS value#7]
+- 'MapElements <function1>, obj#6: java.lang.String
   +- 'DeserializeToObject unresolveddeserializer(createexternalrow(getcolumnbyordinal(0, StringType).toString, StructField(value,StringType,true))), obj#5: org.apache.spark.sql.Row
      +- Filter <function1>.apply
         +- StreamingRelation FileSource[/Users/zsx/stream], [value#0]

== Analyzed Logical Plan ==
value: string
SerializeFromObject [staticinvoke(class org.apache.spark.unsafe.types.UTF8String, StringType, fromString, input[0, java.lang.String, true], true) AS value#7]
+- MapElements <function1>, obj#6: java.lang.String
   +- DeserializeToObject createexternalrow(value#0.toString, StructField(value,StringType,true)), obj#5: org.apache.spark.sql.Row
      +- Filter <function1>.apply
         +- StreamingRelation FileSource[/Users/zsx/stream], [value#0]

== Optimized Logical Plan ==
SerializeFromObject [staticinvoke(class org.apache.spark.unsafe.types.UTF8String, StringType, fromString, input[0, java.lang.String, true], true) AS value#7]
+- MapElements <function1>, obj#6: java.lang.String
   +- DeserializeToObject createexternalrow(value#0.toString, StructField(value,StringType,true)), obj#5: org.apache.spark.sql.Row
      +- Filter <function1>.apply
         +- StreamingRelation FileSource[/Users/zsx/stream], [value#0]

== Physical Plan ==
*SerializeFromObject [staticinvoke(class org.apache.spark.unsafe.types.UTF8String, StringType, fromString, input[0, java.lang.String, true], true) AS value#7]
+- *MapElements <function1>, obj#6: java.lang.String
   +- *DeserializeToObject createexternalrow(value#0.toString, StructField(value,StringType,true)), obj#5: org.apache.spark.sql.Row
      +- *Filter <function1>.apply
         +- StreamingRelation FileSource[/Users/zsx/stream], [value#0]
```

- Add `StreamingQuery.explain` to display the last execution plan. E.g.,
```
== Parsed Logical Plan ==
SerializeFromObject [staticinvoke(class org.apache.spark.unsafe.types.UTF8String, StringType, fromString, input[0, java.lang.String, true], true) AS value#7]
+- MapElements <function1>, obj#6: java.lang.String
   +- DeserializeToObject createexternalrow(value#12.toString, StructField(value,StringType,true)), obj#5: org.apache.spark.sql.Row
      +- Filter <function1>.apply
         +- Relation[value#12] text

== Analyzed Logical Plan ==
value: string
SerializeFromObject [staticinvoke(class org.apache.spark.unsafe.types.UTF8String, StringType, fromString, input[0, java.lang.String, true], true) AS value#7]
+- MapElements <function1>, obj#6: java.lang.String
   +- DeserializeToObject createexternalrow(value#12.toString, StructField(value,StringType,true)), obj#5: org.apache.spark.sql.Row
      +- Filter <function1>.apply
         +- Relation[value#12] text

== Optimized Logical Plan ==
SerializeFromObject [staticinvoke(class org.apache.spark.unsafe.types.UTF8String, StringType, fromString, input[0, java.lang.String, true], true) AS value#7]
+- MapElements <function1>, obj#6: java.lang.String
   +- DeserializeToObject createexternalrow(value#12.toString, StructField(value,StringType,true)), obj#5: org.apache.spark.sql.Row
      +- Filter <function1>.apply
         +- Relation[value#12] text

== Physical Plan ==
*SerializeFromObject [staticinvoke(class org.apache.spark.unsafe.types.UTF8String, StringType, fromString, input[0, java.lang.String, true], true) AS value#7]
+- *MapElements <function1>, obj#6: java.lang.String
   +- *DeserializeToObject createexternalrow(value#12.toString, StructField(value,StringType,true)), obj#5: org.apache.spark.sql.Row
      +- *Filter <function1>.apply
         +- *Scan text [value#12] Format: org.apache.spark.sql.execution.datasources.text.TextFileFormat1836ab91, InputPaths: file:/Users/zsx/stream/a.txt, file:/Users/zsx/stream/b.txt, file:/Users/zsx/stream/c.txt, PushedFilters: [], ReadSchema: struct<value:string>
```

## How was this patch tested?

The added unit tests.

Author: Shixiong Zhu <shixiong@databricks.com>

Closes #13815 from zsxwing/sdf-explain.