The first few weeks of operation the "backup" gateway would suspend your account for not providing the necessary volume of sales.
I can't explain our difference experiences - maybe it's down to providers. We've been through five, and are currently very happy with Stripe.
Then eventually you can make a decision which works best. The new provider (the backup) or the old provider. Then turn the volume to the best one.
Though I would never go 100% on any of them. Maybe keep some geoip through the backup. E.g. South America with one provider. France with another. Or a 5% load balanced volume.
It is good to keep them all up to date and active, as from experience they all frequently f-up and you have to feature toggle a provider off for a day or week now and then, or at least throttle them down.
From my experience you end up being tied pretty hard to pretty one provider and as long as you make fraud prevention yourself you can stay in business. Problem is that if you do all the fraud prevention stuff yourself then why do you need braintree for? To just invoke a .charge(100, "EUR")??
The rate difference would be an investment in redundancy. If Provider A goes down, I would guess you'll be grateful to have Provider B taking the failover traffic, even if costs you an extra 2.9% in transaction costs. You'd keep feeding Provider B the minimally-acceptable amount of traffic at the higher price point to keep that redundancy alive.
I'd need to do more research to find the best solution but it looks like payment processor risk is definitely something that can be mitigated.