Add session/download-pdf endpoints and add graph to represent values

[VirajP1002]

What is the context of this PR?

To update the benchmark outputs in order for us to monitor and highlight the performance of more intensive endpoints by including session and download-pdf.

benchmark_stats has been provisioned with the 99th percentiles for the session and download-pdf endpoints. In order to achieve this, the happy path has been updated to include download-pdf such that the csv file created when running the benchmark contains the percentiles for this endpoint. The output "string" of the benchmark stats has been updated to include the new metrics for both Github and non-Github outputs.

visualise_results has been updated to support the new metrics. Currently, as the 99th percentiles for both session and download-pdf are significantly larger than the original figures, it was decided that outputting 2 graphs was the most appropriate solution. Thus, 2 dataframes have been created, one for the original percentiles and a new one for the additional metrics. The "image" of the graphs created use subplots to place both graphs created next to each other.

Various tests have been updated and parametrised. In addition to this conftest was updated to include more fixtures that support the new endpoints.

How to review

• Check if the changes make sense
• Suggestions for refactoring
• Ensure all tests pass
• Use the README to test the updated visualise_results script
• Spin up a benchmark pipeline and test the output to Slack and spin up a Dependabot staging pipeline to test the output to Github

[petechd]

Do we want to fix the other places where we cast integer on floats and round up/down instead (as part of this PR, since we are already here)? Are there other places that still do that? Do other devs have any thoughts?

[VirajP1002]

Do we want to fix the other places where we cast integer on floats and round up/down instead (as part of this PR, since we are already here)? Are there other places that still do that? Do other devs have any thoughts?

I'm happy to add those in as it removes the rounding issue we talked about. These changes are minor and I guess it would be potentially tweaking the graph values? So I think it's somewhat relevant to this PR, but yep, let's see what the other devs think

[berroar]

We should bump the test coverage as part of this PR as I think we are up to 85% now 🎉