feat(dataset-builder): add retry logic and ASCII-safe logging for Siemens PDF pipeline

- question_generator.py: add max_retries=3/retry_delay=5s loop with
  exponential backoff on LLM timeout or server errors; encode filenames
  with ascii/replace before printing to avoid UnicodeEncodeError on
  Windows cp1252 consoles
- runner.py: encode PDF filenames ASCII-safe for progress messages;
  catch generation failures per-document and skip (or re-raise) based
  on failure_mode, preventing one bad doc from aborting the whole build

Co-Authored-By: Claude Opus 4 <noreply@anthropic.com>

rag_eval/dataset_builder/runner.py

@@ -111,12 +111,32 @@ def run_dataset_build(
             continue
 
         documents.append(document)
-        generated = generator.generate(
-            document,
-            max_questions=job.max_questions_per_document,
-            max_chunks_per_question=job.max_source_chunks_per_question,
-            job_name=job.job_name,
-        )
+        doc_name_safe = pdf_path.name.encode("ascii", "replace").decode("ascii")
+        print(f"  [info] generating questions for: {doc_name_safe}")
+        try:
+            generated = generator.generate(
+                document,
+                max_questions=job.max_questions_per_document,
+                max_chunks_per_question=job.max_source_chunks_per_question,
+                job_name=job.job_name,
+            )
+        except Exception as exc:
+            gen_failure = ParseFailure(file_path=pdf_path.as_posix(), error=f"generation failed: {exc}")
+            failures.append(gen_failure)
+            print(f"  [warn] skipping {doc_name_safe} after generation failure: {exc}")
+            if job.failure_mode == "fail":
+                result = DatasetBuildResult(
+                    job=job,
+                    run_id=run_id,
+                    artifact_paths=artifact_paths,
+                    documents=documents,
+                    draft_samples=draft_samples,
+                    parse_failures=failures,
+                )
+                write_dataset_build_artifacts(result)
+                raise
+            continue
+
         valid_generated = []
         for sample in generated:
             errors = validate_draft_sample(
@@ -126,9 +146,9 @@ def run_dataset_build(
             )
             if not errors:
                 valid_generated.append(sample)
-        draft_samples.extend(
-            dedupe_samples(valid_generated)[: job.max_questions_per_document]
-        )
+        new_samples = dedupe_samples(valid_generated)[: job.max_questions_per_document]
+        draft_samples.extend(new_samples)
+        print(f"  [info] {doc_name_safe}: {len(new_samples)} questions generated (total so far: {len(draft_samples)})")
 
     result = DatasetBuildResult(
         job=job,