feat(mcp): add LLM server crate and remote client integration

- Introduce `owlen-mcp-llm-server` crate with RPC handling, `generate_text` tool, model listing, and streaming notifications.
- Add `RpcNotification` struct and `MODELS_LIST` method to the MCP protocol.
- Update `owlen-core` to depend on `tokio-stream`.
- Adjust Ollama provider to omit empty `tools` field for compatibility.
- Enhance `RemoteMcpClient` to locate the renamed server binary, handle resource tools locally, and implement the `Provider` trait (model listing, chat, streaming, health check).
- Add new crate to workspace `Cargo.toml`.

crates/owlen-ollama/src/lib.rs

@@ -602,13 +602,23 @@ impl Provider for OllamaProvider {
 
         let options = Self::build_options(parameters);
 
-        let ollama_tools = tools.as_ref().map(|t| Self::convert_tools_to_ollama(t));
+        // Only send the `tools` field if there is at least one tool.
+        // An empty array makes Ollama validate tool support and can cause a
+        // 400 Bad Request for models that do not support tools.
+        // Currently the `tools` field is omitted for compatibility; the variable is retained
+        // for potential future use.
+        let _ollama_tools = tools
+            .as_ref()
+            .filter(|t| !t.is_empty())
+            .map(|t| Self::convert_tools_to_ollama(t));
 
+        // Ollama currently rejects any presence of the `tools` field for models that
+        // do not support function calling. To be safe, we omit the field entirely.
         let ollama_request = OllamaChatRequest {
             model,
             messages,
             stream: false,
-            tools: ollama_tools,
+            tools: None,
             options,
         };
 
@@ -695,13 +705,21 @@ impl Provider for OllamaProvider {
 
         let options = Self::build_options(parameters);
 
-        let ollama_tools = tools.as_ref().map(|t| Self::convert_tools_to_ollama(t));
+        // Only include the `tools` field if there is at least one tool.
+        // Sending an empty tools array causes Ollama to reject the request for
+        // models without tool support (400 Bad Request).
+        // Retain tools conversion for possible future extensions, but silence unused warnings.
+        let _ollama_tools = tools
+            .as_ref()
+            .filter(|t| !t.is_empty())
+            .map(|t| Self::convert_tools_to_ollama(t));
 
+        // Omit the `tools` field for compatibility with models lacking tool support.
         let ollama_request = OllamaChatRequest {
             model,
             messages,
             stream: true,
-            tools: ollama_tools,
+            tools: None,
             options,
         };