feat(subsystembenchmarks): print markdown result table

[zhixiangli]

Summary

Prints benchmark results as a Markdown table in shell output upon benchmark completion, improving immediate readability during CLI and Cloud Build runs.

Changes

• Added print_csv_to_shell() helper in _common/report.py using PrettyTable with TableStyle.MARKDOWN.
• Added prettytable==3.17.0 dependency to Hugging Face dataset requirements.
• Updated run.py to print the formatted Markdown results table after benchmark completion and metric scraping.
• Added upload log output in upload-results.sh.

[gemini-code-assist]

Code Review

This pull request adds functionality to print generated CSV benchmark reports as Markdown tables in the shell using prettytable, and adds an upload progress message to the build script. The review feedback highlights that importing prettytable at the top level of the shared report.py module will cause ImportError crashes in benchmark environments where the dependency is not installed. To resolve this, the reviewer suggests importing prettytable locally inside print_csv_to_shell with graceful error handling, wrapping the CSV parsing in a try-except block to prevent benchmark run failures, and using row.get(header, "") to safely handle missing row values.