Parsing of results_path in Evaluate class is strange

[JohnGiorgi]

The results filepath that is automatically determined in the Evaluate class is really brittle

OCW/ocw/evaluate_only_connect.py

Lines 110 to 116 in ff77ce5

	with open(
	self.results_path
	+ self.prediction_file.split("_")[-2].split("/")[-1]
	+ "_results.json",
	"w",
	) as f:
	json.dump(oc_eval_results, f, indent=2)

I am not sure what the original intention was but its makes it difficult to provide a non-default argument for results_path. A better solution would just to allow a user to provide an output JSON filepath directly, or make it explicit they are providing a directory, and come up with a default filename.