Runner: some improvements and fixes

[siddharth-krishna]

• Use argparse to parse CLI args more robustly
• Also save information like timestamp, timeout, machine configuration (numCPUs, memory), hostname to each row of the results CSV, so that when we parallelize running on multiple VMs on the cloud, we can concat the results and still know which benchmark was run where
• Add CLI arg to upload result to GCP buckets/some other location
• Add a timeout column to the result CSV files so that when we run different benchmarks with different TOs, the website can display them and calculate SGM appropriately (see Home: SGM on intersection of benchmarks solved by all #114 (comment))

Kristijan, just assigning you for FYI for now as you'll be working on the runner. But feel free to prioritize and work on whatever makes sense, thanks.