feat(video): route video pipelines through interpolation helper

LTX, video_run, and framepack_worker bypass process_images_inner, so
they call apply_video_interpolation explicitly before save_video. Save
receives already-inflated frames; the sentinel guard skips its own pass.

- LTX and video_run scale mp4_fps by interpolation_factor(p) so duration
  is preserved instead of stretched (LTX is conditioned on source fps)
- FramePack pre-divides at gen time per get_latent_paddings, so save fps
  stays at mp4_fps; worker passes p=None so save call uses
  mp4_interpolate=0 to skip directly
- replaces the inline (mp4_interpolate+1) fps math at LTX with the
  helper-driven equivalent

modules/framepack/framepack_worker.py

@@ -322,6 +322,15 @@ def worker(
                 if is_last_section:
                     break
 
+            if mp4_interpolate > 0:
+                from modules.processing_video import apply_video_interpolation
+                # history_pixels is 5-D (N,C,T,H,W) in [-1,1]; RIFE needs 4-D (T,C,H,W) in [0,1]
+                x = history_pixels.squeeze(0).permute(1, 0, 2, 3)
+                x = (x.clamp(-1., 1.) + 1.0) * 0.5
+                x = apply_video_interpolation(None, x, count=mp4_interpolate)
+                x = x * 2.0 - 1.0
+                history_pixels = x.permute(1, 0, 2, 3).unsqueeze(0)
+
             total_generated_frames, _video_filename, _thumb = save_video(
                 p=None,
                 pixels=history_pixels,
@@ -334,7 +343,7 @@ def worker(
                 mp4_sf=mp4_sf,
                 mp4_video=mp4_video,
                 mp4_frames=mp4_frames,
-                mp4_interpolate=mp4_interpolate,
+                mp4_interpolate=0,
                 pbar=pbar,
                 stream=stream,
                 metadata=metadata,