Allow for binning where z* is not in the center of a bin

[trilnick]

I'm not sure why it was set up this way to begin with. It's possible to change. However, since the calculations are done by binning, a highly asymmetrical placement could give weird results. I wonder if users can be trusted with this. In any case, perhaps I could add the option for users to input their own bins.

[trilnick]

Will need to tackle the issue of zstar between bins

[trilnick]

I though it could be easy-is to let users supply their own bins. But then I have to change a lot of the architecture:

• Finding in which bin zstar is, and which are the other relevant bins.
• Rather than calculating distances using binw, I'd have to use some sort of numbering of histogram bins.
• Calculating probability masses is now done using binw as well, and that would need to change.

For now, it seems like the simplest solution for anyone who wants the kink/notch point to be between bins is to approximate this with narrower bins. This might make a noisier histogram, but perhaps that can be fixed by adding replications of the data with small variations (e.g. "jitter").