Add CostAwareModel mixin and extract_baseline_plan for cost regression testing

[homatthew]

What am I trying to do?

Enable automated detection of unintended cost calculation changes when capacity models are modified.

Why did I do it this way?

• Mixin pattern: Not all models implement cost methods yet; allows incremental adoption
• Two cost methods: cluster_costs() for infrastructure, service_costs() for workload-dependent costs
• Reuses _sub_models() DFS: Same traversal as plan_certain() for consistency

Algorithm

extract_baseline_plan(model_name, desires):
    # 1. Convert CurrentClusters → ClusterCapacity (price instances + drives)
    # 2. DFS cycle check (error if model composes with itself)
    # 3. DFS traverse via _sub_models(), for each model:
    #    - Require CostAwareModel (error if missing)
    #    - Call cluster_costs() + service_costs()
    # 4. Return CapacityPlan with aggregated costs

Architecture

graph LR
    A[Current Deployment] --> B[extract_baseline_plan]
    B --> C[DFS via _sub_models]
    C --> D[Each model: cluster_costs + service_costs]
    D --> E[Aggregated CapacityPlan]
    E --> F[Compare to baseline]

Loading

Tests

Regression tests verify costs stay stable. See tests/netflix/test_cost_regression.py.

🤖 Generated with Claude Code