Cleanup visualisation code prior to 0.1 release

[cjayb]

I strongly suggest we make decisions on these points before releasing. I fear they'll haunt us otherwise!

• to baseline_renormalize or not (a.k.a. postproc=What) (see [WIP] subtract exponential to remove offset (a.k.a baseline_renormalize) #270; deferred to 0.2)
• decide on and implement a principled way to handle units, esp. when plotting (done in [MRG] updated visualisation methods and unit handling #264; units are nAm)

On the first point, I propose changing to postproc=False (and modifying all tests to explicitly set it to True). I think this should be the case regardless of what the default behaviour of HNN GUI needs to be.

On the second, I like @jasmainak 's proposal to keep everything in fAm, and use scalings-arguments to plot in given units. @rythorpe also pointed out that legends can be used to indicate what amount of scaling has been applied.

These seem like two quite straightforward PRs to me, should we include them in 0.1?

[jasmainak]

I am okay with this but we might need a section in the documentation to explain how to get the "old behavior" and explaining to folks what has changed and why. It's partly there in the whats_new but maybe worth pulling it out into a separate document might be helpful. Mostly to get buy-in from folks who have been using HNN-GUI in the past.

For postproc, another option is to set postproc=True but have a deprecation warning that next cycle it will become postproc=False. But probably not worth the effort

[ntolley]

This may not be worth taking the time to implement but the quantities package is definitely aimed at this exact issue of units.

I've seen it implemented in a few neuroscience specific toolkits (neo and elephant specifically), but at the moment it is tough to say how many headaches it would save.