feat(studio): system-ai editor — test-without-save + model picker (v0.13.0)

Three operator-facing improvements driven by Stefan's testing:

  * **Test Connection** runs against the form's *current* values
    and no longer requires saving first. Previously we save-then-
    test which caused a "system AI is disabled" message when the
    form's model field was still empty. Now the test prompt
    travels with the request body so unsaved tweaks can be
    validated.

  * **Model picker** replaces the free-text Model field. A
    Refresh button calls HubAdmin.ListSystemAiModels (provider's
    `/v1/models`); installed models show up as clickable chips
    that fill the field with one tap. The text input stays for
    cases where the operator already knows the id.

  * **Pull** button (Ollama only) downloads a model via
    `/api/pull`. Type `gemma3:4b`, click Pull, spinner, then the
    new model lights up in the chip list. No more "switch to a
    terminal to ollama-pull" detour.

Helper text updates for the empty-model case ("Required: pick
from the list (Refresh) or type one. Use Pull to download from
Ollama.") so a fresh operator never wonders why Save / Test
errors out.

Signed-off-by: flemming-it <stefan.a.flemming@googlemail.com>

lib/data/hub.dart

@@ -163,11 +163,24 @@ class HubService {
     );
   }
 
-  /// Probe the configured provider with a tiny ping. Same
-  /// error-kind taxonomy as [askAi]; "Test connection" buttons
-  /// use this to surface "is it working?".
-  Future<AskAiResult> testSystemAi() async {
-    final r = await _client.testSystemAi();
+  /// Probe a System-AI configuration with a tiny ping. Pass
+  /// the editor's current form values so "Test connection"
+  /// works *before* save. Empty fields fall back to the live
+  /// in-memory config.
+  Future<AskAiResult> testSystemAi({
+    String provider = '',
+    String endpoint = '',
+    String model = '',
+    String apiKeyEnv = '',
+    String privacyMode = '',
+  }) async {
+    final r = await _client.testSystemAi(
+      provider: provider,
+      endpoint: endpoint,
+      model: model,
+      apiKeyEnv: apiKeyEnv,
+      privacyMode: privacyMode,
+    );
     return AskAiResult(
       errorKind: r.errorKind,
       text: r.text,
@@ -175,6 +188,48 @@ class HubService {
     );
   }
 
+  /// List the models the provider exposes via `GET /v1/models`.
+  /// Studio's editor uses this to populate a dropdown instead
+  /// of forcing the operator to type the id manually.
+  Future<({String errorKind, String text, List<String> ids})>
+      listSystemAiModels({
+    String provider = '',
+    String endpoint = '',
+    String apiKeyEnv = '',
+  }) async {
+    final r = await _client.listSystemAiModels(
+      provider: provider,
+      endpoint: endpoint,
+      apiKeyEnv: apiKeyEnv,
+    );
+    return (
+      errorKind: r.errorKind,
+      text: r.text,
+      ids: r.models.map((m) => m.id).toList(),
+    );
+  }
+
+  /// Pull (download) an Ollama model via /api/pull.
+  /// Synchronous; can take minutes for large models. Errors
+  /// come back as `errorKind`.
+  Future<({String errorKind, String text, int elapsedMs})>
+      pullSystemAiModel({
+    required String endpoint,
+    required String model,
+    String apiKeyEnv = '',
+  }) async {
+    final r = await _client.pullSystemAiModel(
+      endpoint: endpoint,
+      model: model,
+      apiKeyEnv: apiKeyEnv,
+    );
+    return (
+      errorKind: r.errorKind,
+      text: r.text,
+      elapsedMs: r.elapsedMs,
+    );
+  }
+
   /// Active channel + per-channel daemon status snapshot.
   Future<ChannelStatusSnapshot> channelStatus() async {
     final r = await _client.channelStatus();