Document dependency of CCIndexSparkJob to Java S3 file system libs

[sebastian-nagel]

(reported by @calee88 in #12)

While CCSparkJob uses the boto3 to read WARC/WAT/WET files, CCIndexSparkJob requires that the Spark installation includes libs/jars to access data on S3 (s3://commoncrawl/). These are usually provided when a Spark is used in a Hadoop cluster (eg. EMR, Spark on Yarn) but may not for any Spark package esp. when running Spark locally (not in a Hadoop cluster). Also add information about

• Hadoop S3 FilesSystem implementations requiring to adapt the schema part of the data URI (s3:// on EMR, s3a:// when using s3a)
• accessing data anonymously (s3a.AnonymousAWSCredentialsProvider).