Spark 3.0 readiness part 1

[Niharikadutta]

This PR adds support for following APIs added in Spark 3.0:

SparkSession APIs

• python def getActiveSession(cls)
• scala def executeCommand(runner: String, command: String, options: Map[String, String]): DataFrame

DataFrame APIs

• scala/python def tail(n: Int): Array[T]
• scala printSchema(level: Int): Unit
• scala def explain(mode: String): Unit
• scala def observe(name: String, expr: Column, exprs: Column*): Dataset[T]

DataFrameStatFunctions APIs

• scala def sampleBy[T](col: Column, fractions: Map[T, Double], seed: Long): DataFrame

This PR addresses part of #633

[imback82]

Can you resolve conflicts? Thanks!

[Niharikadutta]

After merging from master, I can see VectorUdf tests failing in collectToPython with java.lang.IllegalArgumentException. I am looking into it.

[suhsteve]

LGTM.

[imback82]

Few nits, but LGTM, thanks @Niharikadutta!

[imback82]

indentation looks off?

[Niharikadutta]

Changed, please let me know if that looks better!

[suhsteve]

You can try the following if the indentation looks funny.

        public void PrintSchema(int level)
        {
            var schema = (JvmObjectReference)_jvmObject.Invoke("schema");
            Console.WriteLine((string)schema.Invoke("treeString", level));
        }