Fix FIO OpenSearch field-limit explosion

[sravan27]

Fixes #19.

Summary

• Process FIO archives as one summary document per workload via process_multiple(), matching the existing multi-document orchestration pattern.
• Remove nested per-job jobs arrays from FIO summary metrics and time-series point metrics.
• Preserve per-job time-series detail as scalar job_N_* point metrics in the existing zathras-timeseries export path.
• Add schema validation that rejects nested time-series metric values before they can create unbounded OpenSearch mappings.

Validation

• python -m pytest tests/test_processor_parsing.py tests/test_schema.py -q
• python -m pytest -q
• python -m py_compile src/chronicler/processors/fio_processor.py src/chronicler/schema.py tests/test_processor_parsing.py
• git diff --check