[test] small fix to use vocab_size=1024

Author: feice-huang

tests/pipelines/joyimage/test_joyimage_edit.py

@@ -92,11 +92,13 @@ def get_dummy_components(self):
 
         scheduler = FlowMatchEulerDiscreteScheduler()
 
-        text_encoder = Qwen3VLForConditionalGeneration.from_pretrained(tiny_ckpt_id)
         processor = Qwen3VLProcessor.from_pretrained(tiny_ckpt_id)
         processor.image_processor.min_pixels = 4 * 28 * 28
         processor.image_processor.max_pixels = 4 * 28 * 28
 
+        text_encoder = Qwen3VLForConditionalGeneration.from_pretrained(tiny_ckpt_id)
+        text_encoder.resize_token_embeddings(len(processor.tokenizer))
+
         components = {
             "transformer": transformer,
             "vae": vae,