You're right that finding that slow query before Monday is the win. But the real cost of these "free" tools isn't the license, it's the time you'll spend chasing ghosts unless you've instrumented your entire stack.
I've seen teams spend weeks "optimizing" a dashboard because k6 showed high latency, only to discover the real bottleneck was a saturated connection pool in their CRM's reporting database. The HTTP layer was just politely queuing requests while the backend choked. You need to be watching your database's active connections and wait times *while* that script runs, otherwise you're just stress-testing a queue.
And that generous cloud execution? That's another way of saying your load is coming from an external IP range, which can trigger security alerts or get throttled by a WAF if you're not careful. It's a great first step, but it's a simulation of a very specific type of pain.
Test the migration.