feat(pipeline): per-stage max_tokens, LLM provenance, processing log

- PipelineStage now supports max_tokens to override the 4096 default
- SourcePipeline records provider/model on each entity file as HTML comment
- output/processing-log.yaml tracks tokens, cost, duration, retries, errors
- _call_llm returns (content, metadata) for downstream traceability
- _http.py wraps JSON parse errors with body preview for debugging
- infospace.yaml stages: extract/map=6000 tokens, synthesize=3000 tokens

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

examples/infospace-with-history/infospace.yaml

@@ -45,6 +45,7 @@ pipeline:
       output_dir: output/entities
       output_macro: entities
       split_entities: true
+      max_tokens: 6000
       macros:
         extraction_rules: artifacts/guidelines/extraction-rules.md
         vsm_framework: artifacts/vsm-reference/vsm-framework.md
@@ -52,6 +53,7 @@ pipeline:
       template: templates/map-to-vsm.md
       output_dir: output/mappings
       output_macro: mappings
+      max_tokens: 6000
       macros:
         mapping_rules: artifacts/guidelines/mapping-rules.md
         vsm_framework: artifacts/vsm-reference/vsm-framework.md
@@ -59,6 +61,7 @@ pipeline:
       template: templates/synthesize-analysis.md
       output_dir: output/analyses
       output_macro: analysis
+      max_tokens: 3000
       macros:
         vsm_framework: artifacts/vsm-reference/vsm-framework.md
   post_batch: