As in the discussion of #583, we need to allow dynamic loading of the source file to grab the respected anomaly. To do this, we can introduce an argument path to help localize the source file.

default behavior

In the event when no path is defined, the function should retrieve the source file anomalies.csv hosted on s3 bucket which contains the ground truth of the public datasets we have in Orion.

expected behavior

If path is specified, then use that file as the source to load the anomalies from. The source file should be a csv file with the following format:

 	signal 	events
0 	P-1 	[[1331272800, 1335592800], [1382832000, 1389482000]]
1 	S-1 	[[1398168000, 1407823200]]
2 	E-1 	[[1393027200, 1393675200], [1406203200, 1464436800]]
3 	E-2 	[[1405944000, 1436119200]]

where:

• signal column represents the name of the signal such that the function will fetch the anomalies of that particular signal.
• events column that contains a list of tuples with two elements. the tuple contains the start and end timestamps of the anomaly.

after using load_anomalies(signal, path) the resulting dataframe is

 	start 	    end
0 	1331272800, 1335592800
1 	1382832000, 1389482000

where each entry is an anomaly pertaining to the selected signal