Change behavior of the fh argument in the predict_model function in time_series

Currently, if we want to predict 12 periods in the future we have to pass np.arange(1,13). Instead, it should be 12. This behavior is consistent with how it is currently handled in the setup function.

from pycaret.datasets import get_data
data = get_data('airline')

from pycaret.time_series import *
s = setup(data, fold = 5, fh = 12, session_id = 123)

arima = create_model('arima')

predict_model(arima, fh = 12)

Currently, this produces this:

The changed behavior should be predictions for all 12 points instead of just 12th point.