feat: add OpenTelemetry tracing support behind otel feature flag

- Add tracing::instrument spans to executor (workflow.execute),
  host (workflow.start, event.publish, event.process)
- Add otel feature flag to wfe-core and wfe crates
- Add wfe/src/otel.rs helper for OTLP exporter initialization
- Dependencies: tracing-opentelemetry, opentelemetry, opentelemetry_sdk,
  opentelemetry-otlp (all optional behind otel feature)
- Step execution stays at info level, executor internals at debug

wfe-core/Cargo.toml

@@ -8,8 +8,11 @@ description = "Core traits, models, builder, and executor for the WFE workflow e
 [features]
 default = []
 test-support = []
+otel = ["tracing-opentelemetry", "opentelemetry"]
 
 [dependencies]
+tracing-opentelemetry = { workspace = true, optional = true }
+opentelemetry = { workspace = true, optional = true }
 tokio = { workspace = true }
 serde = { workspace = true }
 serde_json = { workspace = true }

wfe-core/src/executor/workflow_executor.rs

@@ -59,6 +59,15 @@ impl WorkflowExecutor {
     /// 6. Check for completion
     /// 7. Persist
     /// 8. Release lock
+    #[tracing::instrument(
+        name = "workflow.execute",
+        skip(self, definition, step_registry),
+        fields(
+            workflow.id = %workflow_id,
+            workflow.definition_id,
+            workflow.status,
+        )
+    )]
     pub async fn execute(
         &self,
         workflow_id: &str,
@@ -96,6 +105,8 @@ impl WorkflowExecutor {
             .get_workflow_instance(workflow_id)
             .await?;
 
+        tracing::Span::current().record("workflow.definition_id", workflow.workflow_definition_id.as_str());
+
         if workflow.status != WorkflowStatus::Runnable {
             debug!(workflow_id, status = ?workflow.status, "Workflow not runnable, skipping");
             return Ok(());
@@ -170,6 +181,15 @@ impl WorkflowExecutor {
             // Now we can mutate again since context is dropped.
             match step_result {
                 Ok(result) => {
+                    let step_status = if result.sleep_for.is_some() {
+                        "sleeping"
+                    } else if result.event_name.is_some() {
+                        "waiting_for_event"
+                    } else {
+                        "completed"
+                    };
+                    tracing::Span::current().record("step.status", step_status);
+
                     info!(
                         workflow_id,
                         step_id,
@@ -202,6 +222,7 @@ impl WorkflowExecutor {
                 Err(e) => {
                     // f. Handle error.
                     let error_msg = e.to_string();
+                    tracing::Span::current().record("step.status", "failed");
                     warn!(workflow_id, step_id, error = %error_msg, "Step execution failed");
 
                     let pointer_id = workflow.execution_pointers[idx].id.clone();
@@ -253,6 +274,8 @@ impl WorkflowExecutor {
             workflow.complete_time = Some(Utc::now());
         }
 
+        tracing::Span::current().record("workflow.status", tracing::field::debug(&workflow.status));
+
         // Determine next_execution.
         let has_active = workflow.execution_pointers.iter().any(|p| p.active);
         if has_active {