[BUG][Spark Connector]When using FullFidelity change feed mode, the spark job failed with NullPointerException

## Issue:
```
Job aborted due to stage failure: Task 3 in stage 22.0 failed 1 times, most recent failure: Lost task 3.0 in stage 22.0 (TID 79) ```
(d154638bc000009.abqsg4x0r5nu5avvstxro4uyoe.dx.internal.cloudapp.net executor driver): java.lang.NullPointerException
at com.azure.cosmos.spark.CosmosRowConverterBase.parseId(CosmosRowConverterBase.scala:708)
at com.azure.cosmos.spark.CosmosRowConverterBase.$anonfun$convertStructToChangeFeedSparkDataTypeV1$1(CosmosRowConverterBase.scala:835)
at scala.collection.TraversableLike.$anonfun$map$1(TraversableLike.scala:286)
at scala.collection.IndexedSeqOptimized.foreach(IndexedSeqOptimized.scala:36)
at scala.collection.IndexedSeqOptimized.foreach$(IndexedSeqOptimized.scala:33)
at scala.collection.mutable.ArrayOps$ofRef.foreach(ArrayOps.scala:198)
at scala.collection.TraversableLike.map(TraversableLike.scala:286)
at scala.collection.TraversableLike.map$(TraversableLike.scala:279)
at scala.collection.mutable.ArrayOps$ofRef.map(ArrayOps.scala:198)
at com.azure.cosmos.spark.CosmosRowConverterBase.convertStructToChangeFeedSparkDataTypeV1(CosmosRowConverterBase.scala:831)
at com.azure.cosmos.spark.CosmosRowConverterBase.fromObjectNodeToChangeFeedRowV1(CosmosRowConverterBase.scala:93)
at com.azure.cosmos.spark.ChangeFeedPartitionReader.com$azure$cosmos$spark$ChangeFeedPartitionReader$$changeFeedItemFactoryMethodV1(ChangeFeedPartitionReader.scala:110)
at com.azure.cosmos.spark.ChangeFeedPartitionReader$$anon$2.deserialize(ChangeFeedPartitionReader.scala:156)
at com.azure.cosmos.CosmosItemSerializer.deserializeSafe(CosmosItemSerializer.java:95)
at com.azure.cosmos.CosmosItemSerializer$1.deserializeSafe(CosmosItemSerializer.java:212)
at com.azure.cosmos.implementation.Utils.parse(Utils.java:645)
at com.azure.cosmos.implementation.RxDocumentServiceResponse.getQueryResponse(RxDocumentServiceResponse.java:188)
```

## Possible root cause:
The issue happens for delete operation, when current and previous node are both being null (whether the previous image will be returned could be controlled by a backend flag config)

## Fix
* Follow up when the previous image will be returned
* Fix in spark connector code to check whether previous image exists, and handle null cases correctly (for example return null for the properties need to be parsed)