mz440: cleanup kaniko dir

[mzihlmann]

Fixes #440

Description
Kaniko is not designed to run multiple times in the same pod, so adding retries around its execution causes hard failures. While Kaniko provides a --cleanup option, it only removes the filesystem created during the build and only on success. This change extends the cleanup to also remove Kaniko’s working directory, which stores intermediate build output, and ensures cleanup happens even when the build fails.

The following new directories are removed:

• DockerfilePath: The dockerfile we're building, it will be overwritten anyways
• KanikoIntermediateStagesDir: Where we store stages as tarball/oci
• BuildContextDir <- this was the root cause of the issue
• KanikoInterStageDepsDir: Files that are copied between stages
• KanikoLayersDir: Where we store tarball/oci snapshots of individial layers
• KanikoSecretsDir: Where secrets are copied to, it would be wrong to reuse them

We specifically do not remove:

• KanikoCacheDir: We might want to reuse cache dirs, but you could argue that this is dangerous too
• KanikoSwapDir: This directory must be empty anyways
• DockerConfigDir: This is not created by kaniko

It's a bit a debate which directories should be reusable, as it depends on whether we rebuild the same dockerfile, same target etc etc. I know that a handful of people build multiple targets in the same pod using this strategy, we eventually want to support that out of the box in a single run.

Submitter Checklist

These are the criteria that every PR should meet, please check them off as you
review them:

• Includes unit tests
• Adds integration tests if needed.

See the contribution guide for more details.

Reviewer Notes

• The code flow looks good.
• Unit tests and or integration tests added.

Release Notes

Describe any changes here so maintainer can include it in the release notes, or delete this block.

Examples of user facing changes:
- kaniko adds a new flag `--registry-repo` to override registry

[mzihlmann]

reproducer repository here https://gitlab.com/martizih/kaniko-retry

[mzihlmann]

The documentation was already unambiguous that the cleanup will happen on failure too, so that change is correct and I would consider implementation so far bogus.