[SQL] Simple framework for debugging query execution

Only records number of tuples and unique dataTypes output right now...

Example:
```scala
scala> import org.apache.spark.sql.execution.debug._
scala> hql("SELECT value FROM src WHERE key > 10").debug(sparkContext)

Results returned: 489
== Project [value#1:0] ==
Tuples output: 489
 value StringType: {java.lang.String}
== Filter (key#0:1 > 10) ==
Tuples output: 489
 value StringType: {java.lang.String}
 key IntegerType: {java.lang.Integer}
== HiveTableScan [value#1,key#0], (MetastoreRelation default, src, None), None ==
Tuples output: 500
 value StringType: {java.lang.String}
 key IntegerType: {java.lang.Integer}
```

Author: Michael Armbrust <michael@databricks.com>

Closes #1005 from marmbrus/debug and squashes the following commits:

dcc3ca6 [Michael Armbrust] Add comments.
c9dded2 [Michael Armbrust] Simple framework for debugging query execution