Currently, WPR archive are specified on a per storyset basis, e.g: each story_set is linked to a json file (let's call this "json archives file") which basically a table that map each story name to a WPR archive file path.

This mechanism is problematic in a couple of ways:
1) It makes sharing stories between different benchmarks harder. This is because a story code is not self-contained: which WPR archive file to used for the story depends on the story_set.

2) It makes any kind of story renaming harder. An operation as simple as renaming a story will require finding all the story set that use that story, then update the name entries of that story in all the corresponding json archives files.