feat(eval): memoir A/B chapter judging and eval-web parity with dialogue

- Judge baseline excerpt and library chapter separately; build_memoir_compare_summary for gate, nine-dim and leaf deltas.

- Memoir SSE chapter payload: baseline_judge, compare_summary, baseline_judge_error.

- MemoirJudgeOutput: loose score coercion and post-validate clamp; memoir judge prompt caps from settings.

- app-eval-web: two-column MemoirScoreCard layout, MemoirCompareSummary, chapter blocks and CSS.

- Add memoir_compare_summary, log_events, celery_log_context, memoir_pipeline_progress; tests and migration 0014.

- Misc: memory/evidence and enrichment paths, task/orchestrator updates, internal-eval docs, env examples.

api/app/agents/memoir/batch_phase1_prep.py

@@ -6,7 +6,7 @@ from __future__ import annotations
 
 import math
 from dataclasses import dataclass
-from typing import Any, Dict, List
+from typing import Any, Callable, Dict, List
 
 from app.agents.memoir.prompts import get_batch_memoir_phase1_prep_prompt
 from app.agents.memoir.schemas import BatchPhase1LLMOutput
@@ -135,7 +135,7 @@ def _run_batch_phase1_prep_chunk_with_bisect(
         if merged.keys() != expected:
             raise ValueError(
                 "batch phase1 chunked bisect merge: segment ids do not match input"
-            )
+            ) from None
         return merged
 
 
@@ -145,6 +145,7 @@ def run_batch_phase1_prep_chunked(
     llm: Any,
     *,
     chunk_size: int,
+    on_chunk: Callable[[int, int], None] | None = None,
 ) -> Dict[str, BatchPhase1SegmentRow]:
     """
     将 segments 按 chunk_size 切片多次调用 Phase1 批处理 LLM，合并 by_id。
@@ -161,13 +162,16 @@ def run_batch_phase1_prep_chunked(
         chunk_idx = i // chunk_size + 1
         sub = segments[i : i + chunk_size]
         logger.info(
-            "event=batch_phase1_chunk chunk_idx={}/{} segment_count={} batch_path=chunked",
+            "event=batch_phase1_chunk chunk_idx={}/{} segment_count={} batch_path=chunked "
+            "msg=Phase1 批处理分块调用",
             chunk_idx,
             total_chunks,
             len(sub),
         )
         part = _run_batch_phase1_prep_chunk_with_bisect(sub, state, llm)
         merged.update(part)
+        if on_chunk is not None:
+            on_chunk(chunk_idx, total_chunks)
     expected = {str(s.id) for s in segments}
     if merged.keys() != expected:
         missing = expected - merged.keys()