NEXT: bringing in evals and metrics

[tomxi]

Now that we have baseline and proposed method for casting to monotonic boundries, let's see what we can do to evaluate them.

Let's bring in the basic versions with mir_eval first.

Let's also implement the components of B-measure now

• Boundary Hit Rate of each level
• Boundary Ranking recall.

With all of this, it's time to explore the score of baseline methods against reference.

The next step would be to see if there are specific parameter regions of the KDE pipeline that's favored by the metrics....