Add a TimeSeriesAnomalyDetector about orion

[ManuelAlvarezC]

Also, a TimeSeriesLoader class to load the data in a format ready to be used by the TimeSeriesAnomalyDetector. The class constructor will have the following arguments:

• dataset_path: The path to the dataset. It will be an CSV file without index nor headers.
• predictions : Number of steps ahead to predict.
• samples: Number of steps composing each training sample.

And will have a single method load(self, target=True) that will load and format the data and return, for a CSV of size (n_rows, n_features) the following items:

• X: numpy.ndarray a 3-d array of shape (training_samples, samples, n_features )
• y: np.ndarray a 2-d array of shape (training_samples, predictions)