fix(report): raise RuntimeError on non-object run.json (fixes #1109)

[vardhans07]

Summary

Fixes #1109.

• Added type validation in read_run_record() (strix/report/writer.py) to raise RuntimeError when run.json contains a valid but non-dict JSON root (such as a list or string).
• Updated test_read_run_record_non_object_raises in tests/test_report_writer.py to expect RuntimeError instead of TypeError, preventing uncaught exceptions during --resume runs.

[greptile-apps]

Too many files changed for review (416 files, 100 file limit).

Bypass the limit by tagging @greptile-apps to review.

[SebTardif]

Summary

The change correctly maps a parseable non-object run.json ([], string, null, bool, number) onto RuntimeError, which _load_resume_state already catches and turns into parser.error. That closes the --resume TypeError hole in issue #1109, and the other caller (ReportState.hydrate_from_run_dir) does not catch TypeError so it does not regress. Dominant risk: this reverts the exact RuntimeError -> TypeError edit from #576 that was made to satisfy pre-commit TRY004, so ruff will fail again unless there is a noqa or the catch site is widened instead.

Issue counts by severity

• bugs: 1
• suggestions: 1
• nits: 1

[SebTardif]

Previous requested changes are addressed on e3fd04c.

• read_run_record still raises TypeError for a non-object root (TRY004 / #576).
• The writer.py comment and extra blank line are gone.
• _load_resume_state catches TypeError so --resume exits through argparse.
• test_resume_non_object_run_json_exits covers a list-root run.json.

This clears the earlier Changes requested review.