Training samples broken with 2025-02-26 commit

[Enyakk]

Re-Tested. It was this commit that broke training samples:
f7e680b

Resulting in error:
`INFO:main:height: 640
INFO:main:width: 512
INFO:main:frame count: 25
INFO:main:sample steps: 25
INFO:main:guidance scale: 6.0
INFO:main:discrete flow shift: 12.0
Sampling timesteps for prompt 1: 0%| | 0/25 [00:00<?, ?it/s]
Traceback (most recent call last): 0%| | 0/25 [00:00<?, ?it/s]
File "D:\ML\SD\trainers\musubi-latest\hv_train_network.py", line 2462, in
trainer.train(args)
File "D:\ML\SD\trainers\musubi-latest\hv_train_network.py", line 1780, in train
sample_images(accelerator, args, None, global_step, vae, transformer, sample_parameters, dit_dtype)
File "D:\ML\SD\trainers\musubi-latest\hv_train_network.py", line 374, in sample_images
sample_image_inference(accelerator, args, transformer, dit_dtype, vae, save_dir, sample_parameter, epoch, steps)

File "D:\ML\SD\trainers\musubi-latest\hv_train_network.py", line 532, in sample_image_inference
if image_latents is not None:
UnboundLocalError: local variable 'image_latents' referenced before assignment

steps: 1%| | 10/1400 [01:46<4:06:22, 10.64s/it, Average key norm=0.000441, Keys Scaled=0, avr_loss=0.165]
Traceback (most recent call last):
File "C:\Program Files\Python310\lib\runpy.py", line 196, in run_module_as_main
return run_code(code, main_globals, None,
File "C:\Program Files\Python310\lib\runpy.py", line 86, in run_code
exec(code, run_globals)
File "D:\ML\SD\trainers\musubi-latest\venv\Scripts\accelerate.exe_main.py", line 7, in
File "D:\ML\SD\trainers\musubi-latest\venv\lib\site-packages\accelerate\commands\accelerate_cli.py", line 48, in main
args.func(args)
File "D:\ML\SD\trainers\musubi-latest\venv\lib\site-packages\accelerate\commands\launch.py", line 1172, in launch_command
simple_launcher(args)
File "D:\ML\SD\trainers\musubi-latest\venv\lib\site-packages\accelerate\commands\launch.py", line 762, in simple_launcher
raise subprocess.CalledProcessError(returncode=process.returncode, cmd=cmd)
subprocess.CalledProcessError: Command '['D:\ML\SD\trainers\musubi-latest\venv\Scripts\python.exe', 'hv_train_network.py', '--config_file', 'D:\ML\SD\workspace\musubi\2025-02-24\test_v23\config.toml', '--dataset_config', 'D:\ML\SD\workspace\musubi\2025-02-24\test_v23\test_mixset_v2_aspect_no2_448.toml', '--sdpa', '--split_attn', '--mixed_precision', 'bf16', '--gradient_checkpointing', '--max_data_loader_n_workers', '8', '--persistent_data_loader_workers', '--output_dir', 'D:\ML\SD\workspace\musubi\2025-02-24\test_v23', '--output_name', '2025-02-27_1805_test_mixset_v2_aspect_no2_448', '--logging_dir', 'D:\ML\SD\workspace\musubi\2025-02-24\test_v23/_logs', '--vae_chunk_size', '32', '--vae_spatial_tile_sample_min_size', '128', '--sample_prompts', 'D:\ML\SD\workspace\musubi\2025-02-24\test_v23\prompt.txt']' returned non-zero exit status 1.`