Ollama max_tokens settings (#17025)

- Support `available_models` for Ollama
- Clamp default max tokens (context length) to 16384.
- Add documentation for ollama context configuration.

crates/language_model/src/settings.rs

@@ -152,6 +152,7 @@ pub struct AnthropicSettingsContentV1 {
 pub struct OllamaSettingsContent {
     pub api_url: Option<String>,
     pub low_speed_timeout_in_seconds: Option<u64>,
+    pub available_models: Option<Vec<provider::ollama::AvailableModel>>,
 }
 
 #[derive(Clone, Debug, Serialize, Deserialize, PartialEq, JsonSchema)]
@@ -276,6 +277,9 @@ impl settings::Settings for AllLanguageModelSettings {
                 anthropic.as_ref().and_then(|s| s.available_models.clone()),
             );
 
+            // Ollama
+            let ollama = value.ollama.clone();
+
             merge(
                 &mut settings.ollama.api_url,
                 value.ollama.as_ref().and_then(|s| s.api_url.clone()),
@@ -288,6 +292,10 @@ impl settings::Settings for AllLanguageModelSettings {
                 settings.ollama.low_speed_timeout =
                     Some(Duration::from_secs(low_speed_timeout_in_seconds));
             }
+            merge(
+                &mut settings.ollama.available_models,
+                ollama.as_ref().and_then(|s| s.available_models.clone()),
+            );
 
             // OpenAI
             let (openai, upgraded) = match value.openai.clone().map(|s| s.upgrade()) {