Today, cincinnati has `graph-builder` and `policy-engine` in the same pod. This results in the policy-engine depending on a single instance of graph-builder to build the upstream graph.

The drawback of this architecture, if the graph-builder fails, the policy-engine graph becomes stale for that specific pod. 

With the `/graph-data` endpoint on graph-builder, we need graph-builder to be resilient while responding to `/graph` requests from policy-engine and `/graph-data` requests externally. 

Definition of Done:

• We need to investigate the impact of moving graph-builder to different pods.
• Create tickets to move graph-builder to different pods.
• Investigate if we'll require an equal number of graph-builder and policy-engine pods after separating them.

Note: 

• Policy-engine already has caching mechanism to cache the upstream graph, this makes sure that we're not bombarding the graph-data with all requests to policy-engine. basically reducing the overall load on the graph-builder.