Upgrade to a recent Hadoop version

[sebastian-nagel]

ia-web-commons is based on a quite old Hadoop version (0.20.2) which should be upgraded.

In production we already use a separate branch hadoop-3.2.2" (I know, branch names should not include a version number) which ia-hadoop-tools depend on.

Maybe it's time to upgrade master as well?

• (plus) having a single branch is easier for users and deployment
• (minus) this would bring ia-web-commons further apart from iipc/webarchive-commons

Detailed upgrades:

• Hadoop 0.20.2-cdh3u4 -> 3.3.5
  • depend on the lean hadoop-client instead of hadoop-core to avoid dependency exclusions
  • use mainline vanilla Hadoop instead of Cloudera libs and remove the Cloudera repository from build configuration (note: the free "Cloudera Distribution of Hadoop (CDH)" was stopped in Oct 2019)
• dependency upgrades (if also required by Hadoop, rely on this version)
  • pig 0.10.0 -> 0.17.0
  • guava 17.0 -> 27.0-jre
  • openjson 1.0.12 -> 1.0.13
  • jsoup 1.15.3 -> 1.16.1
  • dsiutils 2.7.2 -> 2.7.3
  • httpcore 4.4.15 -> 4.4.16
  • joda-time 2.11.2 -> 2.12.5

[jnioche]

Should the file pom-cdh4.xml be removed?