sd/ltx: Make chunked_io a flag in its own right

Taking this bi-direcitonal, so make it a for-purpose named flag.

comfy/ldm/lightricks/vae/causal_video_autoencoder.py

@@ -1122,6 +1122,8 @@ class processor(nn.Module):
         return (x - self.get_buffer("mean-of-means").view(1, -1, 1, 1, 1).to(x)) / self.get_buffer("std-of-means").view(1, -1, 1, 1, 1).to(x)
 
 class VideoVAE(nn.Module):
+    comfy_has_chunked_io = True
+
     def __init__(self, version=0, config=None):
         super().__init__()
 

comfy/sd.py

@@ -953,7 +953,7 @@ class VAE:
 
             # Pre-allocate output for VAEs that support direct buffer writes
             preallocated = False
-            if hasattr(self.first_stage_model, 'decode_output_shape'):
+            if getattr(self.first_stage_model, 'comfy_has_chunked_io', False):
                 pixel_samples = torch.empty(self.first_stage_model.decode_output_shape(samples_in.shape), device=self.output_device, dtype=self.vae_output_dtype())
                 preallocated = True