Small changes to benchmarking script

[remi-or]

This PR:

• adds throughput to the final pretty print at the end of the benchmark runs
• adds the information of the output shape to the decoded text of the output (usefull to double check if throughput is right)
• changes the behavior of the run_benchmarks.py script so that it will only run 3 configs unless instructed to do otherwise

This way, anyone can run a "quick"

python benchmark_v2/run_benchmarks.py --model-id "meta-llama/Meta-Llama-3-8B" -b 32 -s 128 -n 256

after changing the generate code or some model specific code to check perf did not take a big hit.
Thanks @SunMarc for the suggestion!

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[SunMarc]

Thanks a lot !