[Task]: Align Spark/Hadoop versions in Spark multiple-versions unit/VR tests

[aromanenko-dev]

What needs to happen?

Currently, we run Spark runner tests against multiple Spark/Hadoop versions. We need to make sure that these versions are properly aligned and officially compatible among them. Also, we need to make sure that only the tested version of Hadoop/Spark classes is sitting on classpath.

Issue Priority

Priority: 3 (nice-to-have improvement)

Issue Components

• Component: Python SDK
• Component: Java SDK
• Component: Go SDK
• Component: Typescript SDK
• Component: IO connector
• Component: Beam examples
• Component: Beam playground
• Component: Beam katas
• Component: Website
• Component: Spark Runner
• Component: Flink Runner
• Component: Samza Runner
• Component: Twister2 Runner
• Component: Hazelcast Jet Runner
• Component: Google Cloud Dataflow Runner

[aromanenko-dev]

CC: @mosche

[Abacn]

Probably related if find helpful, recently I made a change to regularly publish spark job server dev containers to gcr.io/apache-beam-testing/beam_portability/beam_spark3_job_server (part of #27595), as we're doing for flink job server