[SPARK-20954][SQL] `DESCRIBE [EXTENDED]` result should be compatible with previous Spark

## What changes were proposed in this pull request?

After [SPARK-20067](https://issues.apache.org/jira/browse/SPARK-20067), `DESCRIBE` and `DESCRIBE EXTENDED` shows the following result. This is incompatible with Spark 2.1.1. This PR removes the column header line in case of those command.

**MASTER** and **BRANCH-2.2**
```scala
scala> sql("desc t").show(false)
+----------+---------+-------+
|col_name  |data_type|comment|
+----------+---------+-------+
|# col_name|data_type|comment|
|a         |int      |null   |
+----------+---------+-------+
```

**SPARK 2.1.1** and **this PR**
```scala
scala> sql("desc t").show(false)
+--------+---------+-------+
|col_name|data_type|comment|
+--------+---------+-------+
|a       |int      |null   |
+--------+---------+-------+
```

## How was this patch tested?

Pass the Jenkins with the updated test suites.

Author: Dongjoon Hyun <dongjoon@apache.org>

Closes #18203 from dongjoon-hyun/SPARK-20954.