Remove redundant workflow steps and sleep commands in GCP tests

[teor2345]

Motivation

In PR #7665 I made a minimal changes to fix an urgent CI bug, retrying each Docker command to fix these errors:

• "device or resource busy" error in docker
• "/dev/sdb is apparently in use by the system" error in mke2fs

Specifications

https://docs.docker.com/engine/reference/run/#entrypoint-default-command-to-execute-at-runtime

https://linux.die.net/man/8/mke2fs

Complex Code or Requirements

1. Workflow Cleanups:

• we could replace the second lightwalletd mount with a symlink in entrypoint.sh
• this would let us combine the lightwalletd and non-lightwalletd launch steps to simplify the workflow
• We could cleanup the duplicate commands in entrypoint.sh, but that's already happening in PR ref(docker): combine test and prod entrypoints into one #7660

2. Workflow Wait Times:

• we might want to limit the number of times we retry each command, a step or job timeout would be the simplest way to do this
• we might want to limit the number of recent journal entries that get shown
• some errors might need us to delete incorrectly created volumes or containers before trying again
• some of these delete steps might need to end in || true to ignore any delete failures (if the object wasn't actually created)
• if the retries don't fix the mke2fs errors, calling mke2fs -F -F might (two -Fs will force overwriting busy devices)

Testing

These cleanups don't need to be tested multiple times, but we should set a timeout that's significantly longer than the last few job runs.

[teor2345]

This ticket isn't relevant any more