Remove PMF, fix NaN in gradient visualizations, fix pos_open/neg_open shared param

- Remove PMF from ground classification options (PDAL recommends SMRF over PMF)
- Auto-detection now uses CSF for urban/complex terrain instead of PMF
- Add z_std > 30m heuristic to auto-select CSF for complex terrain
- Fix pos_open/neg_open lambda missing 'shared' parameter (NameError in workers)
- Fix NaN mask not restored in hillshade, slope, aspect, curvature
  (gradient-based products computed on filled DEM lost NaN transparency)
- Add nan_mask parameter to _save_tif for centralized NaN restoration
- DTM TIF kept by default (no longer deleted after WebP conversion)

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

lidar_pipeline/pipeline.py

@@ -78,8 +78,8 @@ VIZ_STEPS = [
     ('curvature',  generate_curvature),
     ('svf',        generate_svf),
     ('lrm',        generate_lrm),
-    ('pos_open',   lambda d, b, v, r: generate_openness(d, b, v, r, positive=True)),
-    ('neg_open',   lambda d, b, v, r: generate_openness(d, b, v, r, positive=False)),
+    ('pos_open',   lambda d, b, v, r, shared=None: generate_openness(d, b, v, r, positive=True, shared=shared)),
+    ('neg_open',   lambda d, b, v, r, shared=None: generate_openness(d, b, v, r, positive=False, shared=shared)),
     ('mslrm',      generate_mslrm),
     ('tpi',        generate_tpi),
     ('sailore',    generate_sailore),
@@ -323,9 +323,6 @@ class LidarArchaeoPipeline:
             t_pdf = time.time() - t4
             logger.info(f"  ✓ Rapport PDF terminé ({t_pdf:.1f}s)")
 
-        # Step 5: Keep DTM TIF for reuse (regenerating WebPs skips classification)
-        # Use --no-keep-tif to delete DTM files after processing
-
         t_total = time.time() - t_start
         logger.info(f"✓ {basename} terminé en {t_total:.1f}s")
         logger.debug(f"  Détails: classification={t_classif:.1f}s, DTM={t_dtm:.1f}s, PDF={t_pdf:.1f}s")