Add transform argument to plots

[amukh18]

Description

This pull request adds a much-needed `transform` argument to the function plot_trace. Fixes #1018

Checklist

• Does the PR follow official
  PR format?
• Has included a sample plot to visually illustrate the changes? (only for plot-related functions)
• Is the new feature properly documented with an example?
• Does the PR include new or updated tests to cover the new feature (using pytest fixture pattern)?
• Is the code style correct (follows pylint and black guidelines)?
• Is the change listed in changelog?

[amukh18]

@ahartikainen as suggested, I set the default of the transform argument to None. But I get the error
TypeError: 'NoneType' object is not callable

[aloctavodia]

By doing this you are basically writing data = None(data), which is not legal Python. Instead you should do something like:

if transform is not None:
    data = transform(data)

[amukh18]

@aloctavodia Sorry for this elementary mistake! I fixed it.

[aloctavodia]

Maybe a better wording will be "Function to transform data (defaults to None i.e. the identity function)" or probably just "Function to transform data (defaults to None)"

[aloctavodia]

@amukh18 Also notices that plot_trace is not the only function needing this new argument. Other functions are plot_forest, plot_pair, plot_posterior, plot_rank, plot_parallel, plot_violin, plot_density, plot_joint.

[aloctavodia]

@amukh18 any update on this?

[amukh18]

@aloctavodia I deeply apologise for the long delay but I was caught up in my coursework and exams. I am relatively free now and have made the necessary commits.

[amukh18]

I have spotted some errors. I will recommit

[amukh18]

My code passed local pylint checks but the Travis CI build failed saying there was a pylint error

[OriolAbril]

Thanks for the PR!

Before merging, could you add an example in plot_violin docstring (like it is done in the other plots) with two graphics? First a general violinplot without much customization and then showcase the usage of transform, something like az.plot_vioin(idata, var_names="tau", transform=np.log) You can do it in another PR if you prefer or open an issue and leave this for somebody else

[amukh18]

@OriolAbril No problem!
And yes, I'll add those examples

[amukh18]

@OriolAbril I have created the pull request and am working on it.

[OriolAbril]

I am merging this so examples can be added on top of this using transform already