Add an alternate spark version of indexwarcsjob (without mrjob)

[jt55401]

This introduces an alternative version of indexwarcsjob.py - which uses cc-pyspark's CCFileProcessorSparkJob (currently in this PR: commoncrawl/cc-pyspark#45)

Rationale:

• no longer uses mrjob, which we favor less now
• less code
• more reliable - the downloading/uploading code and general framework are improved compared to the older job, because it uses the improved methods from cc-pyspark. (This is somewhat untested theory, but, will be tested soon)

Downsides:

• mrjob has a rather robust set of "ops" functions to create clusers and run jobs created with it on local, EMR, google, hadoop, etc. We make no attempt to replace this part of what has been lost.

[sebastian-nagel]

Depends on commoncrawl/cc-pyspark#45.

[jt55401]

This job is not used anymore :(

[jenenglish]

@sebastian-nagel Trying to make sense of the the multiple PRs still hanging around mrjob. Should this one be closed or should we leave open to consider?