feat(fetch-requests): support llama_parse pipeline

- Fix PipelineType union: "llama_parser" → "llama_parse" (matches backend enum)
- Add "llama_extract" to SSEEventStep union
- PipelineStepper: advance Extract stage on llama_extract events

Dropdown/badge rendering is OpenAPI-spec-driven and picks the new value up
automatically; no other changes required.

src/features/fetch-requests/fetch-requests.models.ts

@@ -26,7 +26,7 @@ export interface EmailSource {
   txn_dicts_count?: number;
 }
 
-export type PipelineType = "heuristic" | "llm" | "llama_parser";
+export type PipelineType = "heuristic" | "llm" | "llama_parse";
 
 export interface FetchRequestCreate {
   source: FileSource | EmailSource;
@@ -90,7 +90,7 @@ export interface ResolveAmbiguityPayload {
 export type SSEEventStep =
   | "load_content" | "raw_lines" | "txn_blocks" | "txn_dicts"
   | "resume_extract" | "extract" | "paused" | "complete" | "enrich"
-  | "save_expenses" | "pipeline";
+  | "save_expenses" | "pipeline" | "llama_extract";
 
 export type SSEEventStatus =
   | "started" | "completed" | "skipped" | "paused" | "progress" | "failed";