Produce two artifacts in Spark for scala_2.11 and scala_2.12

[HotSushi]

Changes that this PR brings:

• renames the module "transportable-udfs-spark" to "transportable-udfs-spark_2.11"
• adds a module "transportable-udfs-spark_2.12" which uses same sourceset as _2.11
• renames the module "transportable-udfs-test-spark" to "transportable-udfs-test-spark_2.11"
• adds a module "transportable-udfs-test-spark_2.12" which uses same sourceset as _2.11
• enhances transportable-udfs-plugin to plan producing + testing artifacts for scala_2.12

Outcome of this PR is that:

• artifact for spark 2.x is renamed from "udf-name-spark" to "udf-name-spark_2.11"
• an additional artifact is produced for spark 3.x called "udf-name-spark_2.12"

Testing can be done:

• by building transportable-udfs-examples module
• testing out build/libs/transportable-udfs-example-udfs-spark_2.11.jar in spark2.x environment
• testing out build/libs/transportable-udfs-example-udfs-spark_2.12.jar in spark3.x environment

How it works? Scala plugin's compileScala task automatically configures the classpath required to build scala code. It does so by looking for "scala-library.jar" in all dependencies declared in the compile classpath. If a module declares dependency on all scala_2.12 artifacts, the "scala-library.jar" that gets inferred is for scala_2.12. Similarly for building scala_2.11 artifact, all dependencies should be built against scala_2.11. In this PR we create separate modules for scala_2.11 and scala_2.12 with appropriately versioned dependencies.

cc: @shardulm94 @wmoustafa

[wmoustafa]

It seems travis-ci validation did not work. Could you double check by building the examples module again locally before merging?