The query planner runs a trial period for candidate plans, then caches a winning plan keyed by the query shape. Two queries below produce the same plan-cache key. Why is that a correctness-safe but performance-risky outcome?
// Same plan-cache key (same shape):
db.orders.find({ status: "shipped", region: "us" })
db.orders.find({ status: "pending", region: "eu" })