fix: issue with quantiles and fitted values

[MMenchero]

What happened

When using quantiles and add_history=True in the forecast method at the same time, I'm getting incorrect columns and columns with missing values.

Version

v0.6.1

Reproducible example

import pandas as pd 
from nixtla import NixtlaClient
nixtla_client = NixtlaClient(api_key = "API key")

df = pd.read_csv('https://raw.githubusercontent.com/Nixtla/transfer-learning-time-series/main/datasets/electricity-short.csv')
fc = nixtla_client.forecast(df, h=8, quantiles=[0.8, 0.9], add_history=True)

This returns the following columns:

unique_id	ds	TimeGPT	TimeGPT-hi-60	TimeGPT-hi-80	TimeGPT-lo-60	TimeGPT-lo-80	TimeGPT-q-80	TimeGPT-q-90

The quantile columns contain missing values and columns with "hi" and "lo" shouldn't be included.