fix: updated llamastack dist and fixed relevant model registration (#3639)

the llamastack distribution added an inline embedding model.
This change broke the current registration and a new set of flags
needed to be introduce to differentiate the registered models from
the embedding one.

Signed-off-by: Brian <bmahabir@bu.edu>

Author: bmahabirbu

packages/backend/src/assets/llama-stack-images.json

@@ -1,3 +1,3 @@
 {
-  "default": "ghcr.io/containers/podman-ai-lab-stack:8d6a4a9a7c587c0a8e44703dd750355256e7a796"
+  "default": "ghcr.io/containers/podman-ai-lab-stack:a06f399ebf7cb2645af126da0e84395db9bb0d1a"
 }

packages/backend/src/managers/llama-stack/llamaStackManager.spec.ts

@@ -308,6 +308,8 @@ test('requestcreateLlamaStackContainerss registers all local models', async () =
     'models',
     'register',
     'Model 1',
+    '--provider-id',
+    'podman-ai-lab',
   ]);
   expect(podmanConnection.execute).toHaveBeenCalledWith(expect.anything(), [
     'exec',
@@ -316,6 +318,8 @@ test('requestcreateLlamaStackContainerss registers all local models', async () =
     'models',
     'register',
     'Model 2',
+    '--provider-id',
+    'podman-ai-lab',
   ]);
 });
 

packages/backend/src/managers/llama-stack/llamaStackManager.ts

@@ -501,6 +501,8 @@ export class LlamaStackManager implements Disposable {
             'models',
             'register',
             model.name,
+            '--provider-id',
+            'podman-ai-lab',
           ]);
         },
       );