Option 1

I have a database schema that looks something like this:
https://dbfiddle.uk/JKCTjadD

1. A Task belongs to a Service (like in ECS)
2. A Task can be reassigned to another Service.
3. TaskConfig tracks the reassignment history (and other configs), and the latest record (using inserted_on DESC) is the Service that a Task is currently assigned to.
4. A Service has a unique constraint (service_name, environment). Environment has its own table but illustrated here as an Enum for brevity.
5. A Task also has a unique constraint (task_name, environment).

The Problem

The environment in Task is duplicated/redundant: the Service associated with the Task already has the environment that the Task is in.

Option 2

I decided to move Service_id from TaskConfig into Task:
https://dbfiddle.uk/QB7k9hK3

I can then use the following query to fetch the latest pairs of (task_name, environment_name):

SELECT DISTINCT ON (task_name, environment_name) * FROM TaskConfig JOIN Task ON Task.id = Task_id JOIN Service ON Service.id = Service_id ORDER BY task_name, environment_name, inserted_on DESC 

The problem

1. I can no longer have a unique constraint on (task_name, environment) since environment is not in the Task table.
2. A future developer could mistakenly select (task_name, Service_id) which will not result in unique (task_name, environment) pairs.

Help

How would you propose to resolve the issues above?
If there are no better ways, which of the 2 options above is the better schema?