Add new test result for slow tests that cause failures statistically.

[Sqeaky]

Currently we have some microbenchmarks in various test suites for Mezzanine Packages. This usually pass. But occasionally one will fail if the build system is particularly loaded. If an algorithm is faster 99% of the time and fails because of system load this shouldn't be test blocking failure. That said this information still needs to be captured. So decisions can be made and potential actions can be taken.

To make this possible a new test result level should be added for poorly performing but passing tests. This needs to be higher than warning and lower than success.