Update show_azure.py

[mst-rajatmishra]

Hello, I tried making some improvements in code,

• I used StringIO to buffer the markdown output and only write it to the file once all experiments are processed.
• ThreadPoolExecutor is used to process experiments in parallel, significantly improving performance if there are many experiments.
• Errors are logged using Python's logging module, with different levels for warnings and errors.
• The validate_config() function checks that all required keys exist in the configuration.
• The output table is properly formatted with placeholders for missing values and formatted columns.

Hope this would be Useful Contribution. Thank You, Rajat Mishra