Replace markdown result files with CSV + JSON output

[tbroden84]

WHY are these changes introduced?

Markdown was a poor storage format for benchmark results: parsing required fragile regex, aggregation logic (medians) was duplicated between bash and Python. Moving to CSV + JSON separates raw measurements from metadata, lets aggregate_results.py compute statistics directly, and makes the data easy to process with standard tools.

Reference: #43

WHAT is this pull request doing?

The benchmark scripts (run_multiple_latency_tests.sh, run_multiple_throughput_tests.sh) now write two files instead of one markdown file:

• a CSV with one row per run (raw measurements, no pre-aggregated summaries)
• a JSON file with test configuration metadata (instance type, version, duration, sizes, rate limits)

aggregate_results.py is updated to read these CSV+JSON pairs and compute medians using statistics.median.

HOW was this pull request tested?

Small test run: #50