Invert precedence and display of result types in dashboard color coding of instance results

[echeran]

This issue is a follow up to #357. In #357, we apply a background color to the link of each instance of an executor. Each executor instance therefore represents exercising a specific version of a specific library under test.

For the dashboard to reflect what is actionable, we should highlight the link when the result type is error or failure, and not bother highlighting the link when the result type is known issue, unsupported, or passing. If we have one executor instance with a failure when all others are 100% passing, it will be hard to visually see the signal of actionable information (the presence of a failure, colored red) in the sea of unactionable info (all other executor instances have 100% passing tests, so they're all colored green). In this scenario, not coloring any green links allows the red link to visually stand out and grab attention.

Also, the set of result types for a single test execution could be a mix of different result types. A single color will be used to colorize the link in the dashboard. Instead of colorizing based on how close to passing we are, instead we should colorizing based on how close to failing we are. In other words, the logic should go like this: "if any result type is failure, color the link red, else if any result type is error, color yellow, else don't colorize the link"

[sffc]

I don't have a strong opinion on whether we should color the passing or the failing cases.