Build Timeout plugin currently takes unstable builds into account when computing the average duration of the last N runs.

Because of this, if a test suite fails due to an early abort N times in a row, the average time of the build drops. When you eventually fix the issue, the build actually runs all the tests, so it inevitably gets timed out.

This is seemingly only recoverable by deleting the build history.

I'd like the build timeout plugin to consider only successful results, so that we can be reasonably sure that all tests actually ran when computing the expected time.