[SPARK-12067] [SQL] Fix usage of isnan, isnull, isnotnull of Column and DataFrame

[yanboliang]

This PR including the following changes:

• [SPARK-11947][SQL] Mark deprecated methods with "This will be removed in Spark 2.0." #9930 has deprecated DataFrame.isNaN, DataFrame.isNull and replaced by DataFrame.isnan, DataFrame.isnull, in this PR we changed Column.isNaN to Column.isnan, Column.isNull to Column.isnull, Column.isNotNull to Column.isnotnull.
• Add Column.notnull as alias of Column.isnotnull following the pandas naming convention.
• Add DataFrame.isnotnull and DataFrame.notnull.

Todo:
If this PR has been accepted, I can do the peer renaming for Python and R in a follow-up PR.

cc @rxin @marmbrus @davies

[SparkQA]

Test build #46936 has finished for PR 10056 at commit edf7e7b.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[marmbrus]

I think its fine to be uniform with SQL in the functions that we call with columns as argument, but this seems like here we are breaking user code for no reason. And if we are going to go lowercase then why are these isnull and not is_null?

For things that are just standard methods on objects I don't see a need to break with scala/java conventions. We can have aliases in python to be compatible with pandas if that is the motivation.

/cc @rxin

[rxin]

I left isnull unchanged in my earlier pull request because I felt it was too commonly used, and decided not to change them.

[yanboliang]

@marmbrus @rxin Thanks for your comments. The conclusion is clear, I will close this PR and send follow-up one(#10097) to add alias at python side in order to be compatible with pandas.