Regression run meta information - save to disk

Nobody has claimed this yet.

Once a run completes, there is no machine readable data linking the output of the run jobs with the regression run itself. The closest is the HTML summary report page that contains hyperlinks to the block level test reports.

Ideally we have a meta file in the scratch directory recording a regression run, and a reference to the individual jobs that were run as part of that regression run.

This should be generated as early in the flow as possible, so that the information can be used even if the overall regression run fails. This also opens up the potential for features such as continuing cancelled regression runs. Rerunning just the failed jobs, and other such features.

Start by tracing the regression-run flow and the code that creates its scratch directory and HTML summary report. Determine where run and job information first becomes available, then define the machine-readable metadata needed to link the regression run with its individual jobs. Done means the metadata is written early enough to remain useful when the overall run fails.

Written by the indexing model from the issue text.