Make running KNN benchmarks easier

[shubhamvishu]

Running KNN benchmarks currently isn't as seamless as executing the general benchmarks. There is a KNN-specific task vector.tasks, but it’s not integrated into the main benchmark runs, which creates a disconnect. Here are a couple of ideas to enhance the user experience:

• Integrate/wire the KNN task into the regular benchmark runs, maybe using the same vectors as those on the nightlies?
• Allow user to pass an flag to general benchmarks script to run the KnnGraphTester after all tasks are complete? (when the change is supposed to be KNN specific/affecting)
• More ideas?