Fundamental issue: test environment reproducibility

To reproduce test failure reliably we need to know:
1)rootfs snapshot, usually updated weekly
2)compiler docker image snapshot - usually updated weekly and depends if there is upstream updates
3)other components usually added to rootfs, so they are consistent with rootfs version
Knowing all this parameters and storing images enough time will allow to reproduce test failure/pass in reasonable timeframe (how long they are stored on storage)

But, many tests fetch components from external sources, and many of them wont be consistent with rootfs provided by KernelCI, which means we might face reproducibility issue.
We need to research this topic, review tests and improve test reproducibility where it is required.
Example: https://github.com/kernelci/kernelci-core/pull/2695
We fetch platform test parameters, and we might not know what was version when it fails.

Start by reviewing the test environment inputs described here: the rootfs snapshot, compiler Docker image snapshot, retention period, and tests that fetch external components. Use PR 2695 as a concrete example and inspect how platform test parameters are recorded. Done means identifying reproducibility gaps and defining improvements so failures and passes can be recreated from versioned inputs.

Written by the indexing model from the issue text.