2- Use string service tiers in session protocol (#20971)

## Summary
- break service tier session/op/app-server protocol fields from the
closed enum to string tier ids
- send the service tier string directly through model requests, prewarm,
compaction, memories, and TUI/app-server turn starts
- regenerate app-server protocol JSON/TypeScript schemas, removing the
standalone ServiceTier TS enum

## Verification
- just fmt
- cargo check -p codex-core -p codex-app-server -p codex-tui
- just write-app-server-schema

---------

Co-authored-by: Codex <noreply@openai.com>

codex-rs/core/src/client.rs

@@ -73,7 +73,6 @@ use codex_otel::current_span_w3c_trace_context;
 use codex_protocol::SessionId;
 use codex_protocol::ThreadId;
 use codex_protocol::config_types::ReasoningSummary as ReasoningSummaryConfig;
-use codex_protocol::config_types::ServiceTier;
 use codex_protocol::config_types::Verbosity as VerbosityConfig;
 use codex_protocol::models::ResponseItem;
 use codex_protocol::openai_models::ModelInfo;
@@ -150,7 +149,7 @@ pub(crate) const WEBSOCKET_CONNECT_TIMEOUT: Duration =
 pub(crate) struct CompactConversationRequestSettings {
     pub(crate) effort: Option<ReasoningEffortConfig>,
     pub(crate) summary: ReasoningSummaryConfig,
-    pub(crate) service_tier: Option<ServiceTier>,
+    pub(crate) service_tier: Option<String>,
 }
 
 /// Session-scoped state shared by all [`ModelClient`] clones.
@@ -683,7 +682,7 @@ impl ModelClient {
         model_info: &ModelInfo,
         effort: Option<ReasoningEffortConfig>,
         summary: ReasoningSummaryConfig,
-        service_tier: Option<ServiceTier>,
+        service_tier: Option<String>,
     ) -> Result<ResponsesApiRequest> {
         let instructions = &prompt.base_instructions.text;
         let input = prompt.get_formatted_input();
@@ -722,11 +721,7 @@ impl ModelClient {
             store: provider.is_azure_responses_endpoint(),
             stream: true,
             include,
-            service_tier: match service_tier {
-                Some(ServiceTier::Fast) => Some("priority".to_string()),
-                Some(service_tier) => Some(service_tier.to_string()),
-                None => None,
-            },
+            service_tier,
             prompt_cache_key,
             text,
             client_metadata: Some(HashMap::from([(
@@ -1168,7 +1163,7 @@ impl ModelClientSession {
         session_telemetry: &SessionTelemetry,
         effort: Option<ReasoningEffortConfig>,
         summary: ReasoningSummaryConfig,
-        service_tier: Option<ServiceTier>,
+        service_tier: Option<String>,
         turn_metadata_header: Option<&str>,
         inference_trace: &InferenceTraceContext,
     ) -> Result<ResponseStream> {
@@ -1215,7 +1210,7 @@ impl ModelClientSession {
                 model_info,
                 effort,
                 summary,
-                service_tier,
+                service_tier.clone(),
             )?;
             let inference_trace_attempt = inference_trace.start_attempt();
             inference_trace_attempt.record_started(&request);
@@ -1293,7 +1288,7 @@ impl ModelClientSession {
         session_telemetry: &SessionTelemetry,
         effort: Option<ReasoningEffortConfig>,
         summary: ReasoningSummaryConfig,
-        service_tier: Option<ServiceTier>,
+        service_tier: Option<String>,
         turn_metadata_header: Option<&str>,
         warmup: bool,
         request_trace: Option<W3cTraceContext>,
@@ -1321,7 +1316,7 @@ impl ModelClientSession {
                 model_info,
                 effort,
                 summary,
-                service_tier,
+                service_tier.clone(),
             )?;
             let mut ws_payload = ResponseCreateWsRequest {
                 client_metadata: response_create_client_metadata(
@@ -1453,7 +1448,7 @@ impl ModelClientSession {
         session_telemetry: &SessionTelemetry,
         effort: Option<ReasoningEffortConfig>,
         summary: ReasoningSummaryConfig,
-        service_tier: Option<ServiceTier>,
+        service_tier: Option<String>,
         turn_metadata_header: Option<&str>,
     ) -> Result<()> {
         if !self.client.responses_websocket_enabled() {
@@ -1514,7 +1509,7 @@ impl ModelClientSession {
         session_telemetry: &SessionTelemetry,
         effort: Option<ReasoningEffortConfig>,
         summary: ReasoningSummaryConfig,
-        service_tier: Option<ServiceTier>,
+        service_tier: Option<String>,
         turn_metadata_header: Option<&str>,
         inference_trace: &InferenceTraceContext,
     ) -> Result<ResponseStream> {
@@ -1530,7 +1525,7 @@ impl ModelClientSession {
                             session_telemetry,
                             effort,
                             summary,
-                            service_tier,
+                            service_tier.clone(),
                             turn_metadata_header,
                             /*warmup*/ false,
                             request_trace,