Refactor BasePlotter's wrapping of Renderer

[banesullivan]

Close #470 and follow up #536 (comment)

This uses functools.wraps to properly wrap the methods on BasePlotter that call to the active Renderer, minimizing duplicated docstrings and argument listing. Note that properties still aren't handled via wrapping but are all explicitly defined in BasePlotter.

I also went ahead and deprecated the loc argument that was in a few places across the BasePlotter API as it was mostly unused. This deprecation was done in favor of having the user activate a given plotter by the subplot method then calling the method of choice. See #536 (comment). I think that using one active renderer managed by subplot() is more in line with our mission to follow Matplotlib conventions and simpler (having loc argument just makes things messy and hard to manage). Also, this is in line with the Zen of Python:

There should be one-- and preferably only one --obvious way to do it.

I also reorganized the Plotter and Renderer class by keeping all of the wrapped methods from Renderer and properties in their respective sections of the class definition.

[akaszynski]

Excellent, this is just want we need to reduce the complexity of BasePlotter. Thanks for wrapping it, I see no problems with the code, and removing the loc kwarg is a good move as it appears to be unused.

[banesullivan]

@GuillaumeFavelier, I did a quick look through mne-python and it looks like you all never used the loc argument but just the subplot() method so all should be good. Either way, it would be great to have your eyes on this too

[GuillaumeFavelier]

LGTM. Great work refactoring all this @banesullivan. Indeed we choose to use subplot() instead of loc in mne-python 👍