Most names of generated_holidays contain spaces, and so fail to plot when passed to plot_forecast_component(). The tidyeval approach is to forego using aes_() or aes_string() and instead use quasiquotation in aes.

This is the same issue as documented [here](https://stackoverflow.com/questions/51658629/ggplot-aes-string-

Is your feature request related to a problem? Please describe.
Simple auto-regressive forecaster with automatic estimation of number of lags.

Describe the solution you'd like
There is a open PR #411 which implements the basic structure but requires some more work to finish it.

Description

Currently, DistributionOutput classes produce a TransformedDistribution when the distribution method is invoked with loc and scale. However, we cannot get mean and variance out of such object:

import mxnet as mx
from gluonts.mx.distribution import GaussianOutput

distr_output = GaussianOutput()

args = mx.nd.array([0.0]), mx.nd.array([1.0])
loc = mx.

Is your feature request related to a current problem? Please describe.
In order to create an outlier detection with Prophet, i need the full dataframe that's return Prophet

Describe proposed solution
Remove the hardcoded ["yhat"] from Prophet.predict add a variable asking to return just yhat or all the predictions: 'yhat_lower', 'yhat_upper', etc..
https://unit8co.github.io/d