Include KernelCI config version information in test data

One of the variables that take part in a test run, besides the kernel version, config applied, compiler environment, etc. is the information about the KernelCI version used to build the kernel, generate the job and run it, and this has always been ignored.

There are two ways, I think, in which KernelCI impacts a test:
1. The kernel config fragments used in the kernel build, if any
2. The way the job is generated and the involved templates

A change in any of these may affect the result of a test, but currently there's no way to tell which KernelCI version was used to run a particular test. We can get the full kernel config used for a build, and that will include all the KernelCI fragments used, but this doesn't make it easier to track down potential test failures introduced by KernelCI changes, as happened in the past.

There were some cases where KernelCI reported regressions that were found to be caused exclusively by changes in the KernelCI kernel config fragments, but there were no way to tell that from the test results, and no tests in place to check for failures caused by these particular changes.

No files, tests, or entry points are named. Start by tracing how test-run data records kernel and configuration inputs, then identify where the KernelCI version used for the build and job generation can be retained and exposed in results. Done means a test result identifies the relevant KernelCI version and supports investigating regressions caused by KernelCI changes.

Written by the indexing model from the issue text.