[SPARK-18295][SQL] Make to_json function null safe (matching it to from_json)

## What changes were proposed in this pull request?

This PR proposes to match up the behaviour of `to_json` to `from_json` function for null-safety.

Currently, it throws `NullPointException` but this PR fixes this to produce `null` instead.

with the data below:

```scala
import spark.implicits._

val df = Seq(Some(Tuple1(Tuple1(1))), None).toDF("a")
df.show()
```

```
+----+
|   a|
+----+
| [1]|
|null|
+----+
```

the codes below

```scala
import org.apache.spark.sql.functions._

df.select(to_json($"a")).show()
```

produces..

**Before**

throws `NullPointException` as below:

```
java.lang.NullPointerException
  at org.apache.spark.sql.catalyst.json.JacksonGenerator.org$apache$spark$sql$catalyst$json$JacksonGenerator$$writeFields(JacksonGenerator.scala:138)
  at org.apache.spark.sql.catalyst.json.JacksonGenerator$$anonfun$write$1.apply$mcV$sp(JacksonGenerator.scala:194)
  at org.apache.spark.sql.catalyst.json.JacksonGenerator.org$apache$spark$sql$catalyst$json$JacksonGenerator$$writeObject(JacksonGenerator.scala:131)
  at org.apache.spark.sql.catalyst.json.JacksonGenerator.write(JacksonGenerator.scala:193)
  at org.apache.spark.sql.catalyst.expressions.StructToJson.eval(jsonExpressions.scala:544)
  at org.apache.spark.sql.catalyst.expressions.Alias.eval(namedExpressions.scala:142)
  at org.apache.spark.sql.catalyst.expressions.InterpretedProjection.apply(Projection.scala:48)
  at org.apache.spark.sql.catalyst.expressions.InterpretedProjection.apply(Projection.scala:30)
  at scala.collection.TraversableLike$$anonfun$map$1.apply(TraversableLike.scala:234)
```

**After**

```
+---------------+
|structtojson(a)|
+---------------+
|       {"_1":1}|
|           null|
+---------------+
```

## How was this patch tested?

Unit test in `JsonExpressionsSuite.scala` and `JsonFunctionsSuite.scala`.

Author: hyukjinkwon <gurwls223@gmail.com>

Closes #15792 from HyukjinKwon/SPARK-18295.