New Feature：New MergeOperator for upsert to ignore null fields by default

[moresun]

Here is a table with columns(A B C D ) ， Column A is primary key ， All fields are type string.
Spark reads a batch of json data from kafka and converts to Dataframe to upsert. But Dataframe may contains "null" or null for some cells because of some missing fields in kafka(.e.g {A:A1,B:B4,C:C4,D:D4}{A:A2,C:C5}{A:A3,B:B5,D:D5}) which could lead to unexpected result in new table.
Since LakeSoul has MergeOperator feature, we could create a new mergeoprator to deal with this case by ignoring null values