In cases like this, a lot depends on the individual Terraform provider implementation, and the API it is talking to.
I had a quick look at the provider source code, and it looks like it is just passing the services list through, maintaining order, to the DigitalOcean API, in a single API call.
At this point, the problem becomes less about how Terraform will behave, and more about how the DigitalOcean API will behave. The services field in the API appears to be a list … but does its concept of identity, for the purposes of applying updates, rely on the ordering of the list? Or on the name field, treating the list somewhat like a map? I do not know the answer to this.
If it turns out the API is happy for you to rearrange the order without downtime, you can just go ahead and apply your original Terraform plan. The plan as visualised by Terraform will appear misleading - but the plan is just Terraform’s attempt to visualise the difference between the start and end state, and it does not always pick the way that actually describes how the ultimate backend API will behave.
If, on the other hand, it turns out the API is going to undeploy and deploy your services if you rearrange the order, you’re a bit stuck, and will have to decide whether that’s worth it, to achieve consistency.