Back-port JDBC column types to 1.x branch.

Author: Marc-Andre Tremblay <marcandre.tr@gmail.com>

Closes #247 from nrstott/feature/custom-jdbc-column-types.

Author: Mathias Bogaert <mathias.bogaert@gmail.com>

Closes #251 from analytically/patch-1.

This patch updates `RedshiftJDBCWrapper.getDriverClass` to automatically recognize the Redshift JDBC 4.2 driver.

Fixes #258.

Author: Travis Crawford <traviscrawford@gmail.com>

Closes #259 from traviscrawford/travis/jdbc42.

We've seen a lot of messages lately regarding the "Invalid S3 URI: hostname does not appear to be a valid S3 endpoint" exception and so thought we should contribute our two cents and the code changes that worked for us. We've tried many approaches listed in that thread including using `spark.executor.extraClassPath` and `spark.driver.extraClassPath` environment variables to prepend to the classpath, including it in the assembled jar or as a shaded jar, Unfortunately many of these approaches failed mainly because we have on the machines themselves the older aws-java-sdk jar and that usually takes precedence. We ended up going with what Josh mentioned earlier about changing the S3 url in the spark-redshift code to add the endpoint to the host (`*.s3.amazonaws.com`).

This logic will try to instantiate a new AmazonS3URI and if it fails, it'll try to add the default S3 Amazon domain to the host.

Author: James Hou <jameshou@data101.udemy.com>
Author: James Hou <james.hou@gmail.com>
Author: Josh Rosen <joshrosen@databricks.com>

Closes #254 from jameshou/feature/add-s3-full-endpoint-v1.

This patch fixes a bug which caused `spark-redshift` to throw `NumberFormatException` when reading NaN or Infinity from Redshift.

This patch fixes the bug by adding special-case handling of the string constants `nan`, `inf`, and `-inf`, which are the values sent back by Redshift during unloads. Note that we still do not support loads of `NaN` to Redshift since Redshift itself does not seem to support this yet (https://forums.aws.amazon.com/thread.jspa?threadID=236367).

Fixes #261.

Author: Josh Rosen <joshrosen@databricks.com>

Closes #269 from JoshRosen/fix-nan.

Added sample code for reading a Redshift table with SparkR.

Author: Ganesh Chand <ganeshchand@gmail.com>
Author: Josh Rosen <joshrosen@databricks.com>

Closes #282 from ganeshchand/patch-1.

…acktrace

Exceptions thrown from Scala's `Await.result` don't include the waiting thread's stacktrace, making it hard to figure out where errors occur. Similar to the fix implemented in Spark in apache/spark#12433, this patch modifies our `Await.result` usages to wrap and rethrow exceptions to capture the calling thread's stack.

Author: Josh Rosen <joshrosen@databricks.com>

Closes #299 from JoshRosen/better-error-reporting.

(cherry picked from commit b4c6053)
Signed-off-by: Josh Rosen <joshrosen@databricks.com>