fix: aggregate probes details in digest

[parkanzky]

When aggregating jsonl reports from multiple garak invocations using garak.analyze.aggregate_reports, probe specs in the start_run and digest entries were not correctly aggregated. This change aggregates the probe specs from each jsonl output.

Tests had included the incorrect digest aggregation and ignored incorrect start_run aggregation. This updates the test and tests both places.

Verification

List the steps needed to make sure this thing works

• Run the tests and ensure they pass python -m pytest tests/
• Run garak several times with different probespecs
• Aggregate output jsonls - python -m garak.analyze.aggregate_reports -o ./aggregated.jsonl <list of report jsonls>
• Verify that the start_run entry plugins.probe_spec string value is a comma separated list of aggregated probespecs which includes all probespecs from all aggregated report jsonls.
• Verify that the final digest entry probespec string value is the same comma separated list of aggregated probespecs
• Generate the HTML report from the aggregated jsonl - python -m garak.analyze.report_digest -w -r ./aggregated.jsonl > ./report.html
• Verify that the HTML config dropdown probe spec: value includes the aggregated probespecs from individual runs.

[github-actions]

DCO Assistant Lite bot All contributors have signed the DCO ✍️ ✅

[parkanzky]

recheck

[parkanzky]

I have read the DCO Document and I hereby sign the DCO

[parkanzky]

recheck

[jmartin-tech]

As this revision now builds the digest instead of making an attempt to merge all the various entires that might exist it results in calculations for z_scores and comments that based on current the installed version's calibration data. I am good with that change as aggregation should be performed using the same version that generated the individual reports, @leondz can you weigh in on if this is an acceptable change.

Changes accounted for, further review in progress.