Fix TypeError in LongCatImageEditPipeline truncation warning

`_encode_prompt` in `LongCatImageEditPipeline` calls `len()` twice on
`all_tokens` when logging the truncation warning:

    f" {self.tokenizer_max_length} input token nums : {len(len(all_tokens))}"

`len(all_tokens)` already returns an `int`, so the outer `len()` raises
`TypeError: object of type 'int' has no len()`. The failure triggers
exactly in the only branch this warning exists for (prompts longer
than `tokenizer_max_length`, default 512), turning the intended
informational warning into a hard crash.

The sibling `LongCatImagePipeline._encode_prompt` has the correct
`{len(all_tokens)}` at line 291, so this is a typo local to the edit
pipeline. Minimal fix: drop the extra `len()` call.

Reproduces with any prompt whose tokenization exceeds 512 tokens:

    pipe = LongCatImageEditPipeline.from_pretrained(...)
    pipe(prompt="a very long prompt ..." * 200, image=...)
    # TypeError: object of type 'int' has no len()

Author: Ricardo-M-L

src/diffusers/pipelines/longcat_image/pipeline_longcat_image_edit.py

@@ -284,7 +284,7 @@ def _encode_prompt(self, prompt, image):
         if len(all_tokens) > self.tokenizer_max_length:
             logger.warning(
                 "Your input was truncated because `max_sequence_length` is set to "
-                f" {self.tokenizer_max_length} input token nums : {len(len(all_tokens))}"
+                f" {self.tokenizer_max_length} input token nums : {len(all_tokens)}"
             )
             all_tokens = all_tokens[: self.tokenizer_max_length]