Improve verbiage

[mikebentley15]

This is related to issue #95. In this project, we use terms like "ground-truth", "good", "bad", "trouble", etc. These adjectives imply something about the correctness of the result. However, FLiT never asserts to be able to confirm or deny the correctness of a result. Instead, FLiT can talk about the reproducibility of a result and if we can get the same or a different result than the "baseline" compilation. I think the term "baseline" is better than "ground-truth".

Some of this can be addressed when dealing with issue #172, to change "good" and "bad" to "same" and "different by ..." respectively, and specify, at least to a few digits, how different the results actually are. This implies reproducibility or the lack thereof, and says nothing of the correctness of either result, because it is totally possible and utterly plausible that some result-altering compiler optimizations actually generate more correct answers (such as FMA or SIMD vectorization for loops). It is a much more complicated question to determine which answer is closer to the "true" answer.