Thank you for the kind words!
Yes, p is the observation/data point being scored as an outlier. P is scored relative to its nearest neighbor -- a set of data points referred to as o. Perhaps "o" should have a subscript in the equation for clarity. Elements "o" are drawn from the k-nearest neighbors of P.
Put another way, reachability distance is a type of distance between two points, p and o. It is not symmetric (reach-dist(p, o) != reach-dist(o, p)).
I hope this helps clarify your understanding.