[SPARK-25571][SQL] Add withColumnsRenamed method to Dataset

[cryeo]

What changes were proposed in this pull request?

Add withColumnsRenamed method to rename multiple columns at a time as follows.

// before
ds.withColumnRenamed("first_name", "firstName")
  .withColumnRenamed("last_name", "lastName")
  .withColumnRenamed("postal_code", "postalCode")

// after
ds.withColumnsRenamed(
  "first_name" -> "firstName",
  "last_name" -> "lastName",
  "postal_code" -> "postalCode"
)
// or
ds.withColumnsRenamed(Map(
  "first_name" -> "firstName",
  "last_name" -> "lastName",
  "postal_code" -> "postalCode"
))

How was this patch tested?

This patch is tested by unit test.

[AmplabJenkins]

Can one of the admins verify this patch?

[wangyum]

These changes seem to be duplicated: https://github.com/apache/spark/pull/22428/files#diff-7a46f10c3cedbf013cf255564d9483cdR2316

[HyukjinKwon]

// before
ds.withColumnRenamed("first_name", "firstName")
  .withColumnRenamed("last_name", "lastName")
  .withColumnRenamed("postal_code", "postalCode")

// after
ds.withColumnsRenamed(
  "first_name" -> "firstName",
  "last_name" -> "lastName",
  "postal_code" -> "postalCode"
)
// or
ds.withColumnsRenamed(Map(
  "first_name" -> "firstName",
  "last_name" -> "lastName",
  "postal_code" -> "postalCode"
))

This doesn't looks useful or making the codes even shorter. I don't think we should add this.

[HyukjinKwon]

Please take a look for PRs or JIRAs before creating them. It's an exact duplicate of SPARK-25430.