At this point, after a century of work, we have what seems to be a compelling performance practice that matches the evidence we can infer from the period.
I wouldn't pay much attention to either Bernstein or Gould on the question of Bach performance practice. I spent years studying this, and Bach recordings by either man sound bizarrely wrong, like someone weeping, emoting, and shouting their way through a recitation of Frost's 'The Road Not Taken'.
> there aren't many clues left in the score, leaving interpretation of the music on the conductor / musicians.
That's both true and false. It's true in that there are few explicit markings like you find on a modern score, and there was an expectation that the performer would take the score and improvise and ornament it according to a commonly shared performance practice. Think of it more like how a really skilled rock band will take a song as a base structure and fill in solos, bridges, ornaments and the like.
It's false in that the musical language was so full of idiom and convention that, if you know how to read it, the clues are all over the place. Especially in Bach where the idioms interlock in an almost crystalline way. There are only a few passages in Bach where I know of two quite different overall interpretations that are both reasonable. Compare Anthony Newman and Wanda Landowska playing the first variation of the Goldberg Variations, about halfway through, the cascading down and up. Newman follows the base line in making it a series of repeats, Landowska carries it through as dips in the same long line. Both work, but it's hard to imagine a third that doesn't fight the structure. Gould's is a ragtime disaster.
Look up "Baroque performance practice" and you'll find a mass of material.