Add a num_of_features argument to h2o.varimp_heatmap function
We forgot to add a limit on the number of variables that are displayed by default in h2o.varimp_heatmap() Let's add a new param called num_of_features since that's the name of the param used in h2o.varimp_plot(). The default there is 10 -- but maybe a good default in the heatmap is 20, since that's the number of models we display by default, and so then the heatmap will have nice square boxes(?). We can play around with it and see what looks best.
In R’s h2o.varimp_plot(), we have the default set to NULL in the function (and it will show at most 10 by default). That might be because someone thought that putting = 10 would be misleading (if there are less than 10 features), but I think it’s actually better to have the integer hardcoded as default. We set the default for top_n in h2o.varimp_heatmap() to 20, so for consistency let’s use an integer instead of NULL. We could also consider changing the default from NULL to 10 in h2o.varimp_plot() too….
This should be added to both R and Python.