feat(cortex-gateway): C3 — propagate vision capabilities through /v1/models

ModelEntry and CortexModelEntry gain a `capabilities: Vec<String>`
field (serde-default for back-compat). The poller copies it verbatim
from each neuron's ModelInfo.capabilities; list_models computes the
union across every node where a model is loaded so a checkpoint loaded
text-only on one neuron and text+vision on another reports both to the
fleet. Catalogue-only and mid-prewarm entries default to empty until
the catalogue gains a capabilities declaration.

Aliases inherit their target's capability union. New gateway test mocks
two nodes with differing capability arrays and asserts the unioned
/v1/models response.

Closes part of #16 (Stage C3).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

crates/cortex-gateway/src/handlers.rs

@@ -414,6 +414,9 @@ async fn list_models(State(fleet): State<Arc<CortexState>>) -> Json<Value> {
                 loaded: false,
                 feasible_on,
                 locations: Vec::new(),
+                // Catalogue profiles don't declare capabilities yet;
+                // the union is filled in Pass 2 from loaded locations.
+                capabilities: Vec::new(),
             },
         );
     }
@@ -438,6 +441,14 @@ async fn list_models(State(fleet): State<Arc<CortexState>>) -> Json<Value> {
                     if was_loaded {
                         e.loaded = true;
                     }
+                    // Union the per-node capabilities so a model loaded
+                    // on several neurons reports every modality any of
+                    // them advertises.
+                    for cap in &entry.capabilities {
+                        if !e.capabilities.contains(cap) {
+                            e.capabilities.push(cap.clone());
+                        }
+                    }
                 })
                 .or_insert_with(|| CortexModelEntry {
                     id: model_id.clone(),
@@ -449,6 +460,7 @@ async fn list_models(State(fleet): State<Arc<CortexState>>) -> Json<Value> {
                     // feasibility; leave empty.
                     feasible_on: Vec::new(),
                     locations: vec![location],
+                    capabilities: entry.capabilities.clone(),
                 });
         }
     }
@@ -498,6 +510,9 @@ async fn list_models(State(fleet): State<Arc<CortexState>>) -> Json<Value> {
                     loaded: false,
                     feasible_on: Vec::new(),
                     locations: vec![location],
+                    // A model that's only mid-prewarm has no loaded
+                    // location to read capabilities from yet.
+                    capabilities: Vec::new(),
                 });
         }
     }
@@ -527,6 +542,7 @@ async fn list_models(State(fleet): State<Arc<CortexState>>) -> Json<Value> {
                 loaded: target_entry.loaded,
                 feasible_on: target_entry.feasible_on,
                 locations: target_entry.locations,
+                capabilities: target_entry.capabilities,
             },
         );
     }

crates/cortex-gateway/src/poller.rs

@@ -107,12 +107,14 @@ async fn poll_neuron(fleet: &CortexState, name: &str, endpoint: &str) {
                             .and_modify(|e| {
                                 e.status = status;
                                 e.vram_estimate_mb = upstream.vram_used_mb;
+                                e.capabilities = upstream.capabilities.clone();
                             })
                             .or_insert_with(|| ModelEntry {
                                 id: upstream.id.clone(),
                                 status,
                                 last_accessed: None,
                                 vram_estimate_mb: upstream.vram_used_mb,
+                                capabilities: upstream.capabilities.clone(),
                             });
                     }
 

crates/cortex-gateway/src/router.rs

@@ -244,6 +244,7 @@ async fn cold_load(
                     status: ModelStatus::Loaded,
                     last_accessed: Some(chrono::Utc::now()),
                     vram_estimate_mb: profile.vram_mb,
+                    capabilities: Vec::new(),
                 },
             );
         }