feat(connectors): add Ollama and LM Studio as a local auth kind

Both servers expose OpenAI-compatible /v1/chat/completions and /v1/models,
so we route them through pi-ai's `openai-completions` API the same way
Fireworks does — but with a per-installation base URL the user enters
during onboarding instead of a hardcoded one. The shared schemas grow a
`local` credential payload carrying just the baseUrl; the credential
resolver surfaces it on `ResolvedCredential.baseUrl`, the runtime bridge
forwards it through `prepareRun`, and the worker plumbs it into
`resolveModel` so the streamed Model<openai-completions> targets the
user's server. The local-cost path short-circuits to zero in cost-writer
since users bill their own hardware, and the live model list is fetched
from the saved baseUrl by `fetchLocalConnectorModels` (Ollama and LM
Studio both speak the standard OpenAI catalog shape). Onboarding ships
a new LocalProviderCard and the Settings panel a LocalForm; both default
to the catalog's per-connector localhost URL and accept any custom URL.

src/main/worker/runtime/index.ts

@@ -19,6 +19,8 @@ interface PrepareRunResult {
   provider: string;
   apiKey: string;
   extraEnv: Record<string, string>;
+  /** OpenAI-compatible server base URL for local connectors (Ollama, LM Studio). */
+  baseUrl: string | null;
   priorMessages: AgentMessage[];
 }
 
@@ -133,7 +135,9 @@ export class WorkerRunner {
       const runEnv = mergeRunEnvs(job.env, prepared.extraEnv);
       redact = buildRedactor(runEnv);
 
-      const model = resolveModel(prepared.provider, job.model);
+      const model = resolveModel(prepared.provider, job.model, {
+        baseUrl: prepared.baseUrl ?? undefined,
+      });
       if (!model) {
         throw new Error(`Unknown model ${job.model} for provider ${prepared.provider}.`);
       }