You have 2 separate issues.  They MAY be related but they may not.

Tableau will use all available resources to do its work.  Seeing high memory or CPU utilization is not in and of itself a problem.  It is only a problem if:

1) You see degradation of the user experience IN PRODUCTION environments (stress testing does not count)

2) The amount of "work" Tableau is being asked to do is beyond the capabilities of the software (this MAY be your issue)

Have you explored the server logs with Tableau's own tools to see what's going on?  Is the usage scenario you are testing an accurate representation of what you can expect in production?  Or are you just trying to see what load makes Tableau break?

You need to turn on Performance Recording on Tableau Server and look at the output after you run your tests to see what exactly is putting so much load on the server.

You know your users and their potential usage patterns, but what you are describing does not sound like a typical production scenario.  I would have to know a LOT more about how you are testing (are you running a continuous loop where a simulated user opens all possible dashboards in a specific order, then incrementing the user count - so that you have multiple users doing everything that it is possible to do with your dashboards, all at the same time, in a fast-running looping process?) to give you any useful advice.

The rules about designing efficient and performant dashboards still apply: what you have described with icons and KPIs doesn't tell me anything about data sources, number of fields, filters, LOD calculations, number of vizzes on a dashboard, size of datasets, and so on.  Those are all things that can have a huge impact on performance.  That is why it is so important to look at the Performance reports to see where the bottlenecks are.

While I am not familiar with Protractor as a testing tool, a quick survey tells me that it might not be the best choice for performance or stress testing of Tableau either.