feat: Add Pi0.5 inference support with PaliGemmaWithExpert architecture

Core Model Implementation:
- Add PI0_5_Policy model with 32-dimensional action space support
- Implement discrete state input processing for robotics tasks
- Add PaliGemmaWithExpert backbone with AdaRMSNorm for flow matching
- Support 16-step action prediction with temporal modeling

Inference Configuration:
- Add comprehensive inference configuration templates
- Support both high-level and detailed inference configs
- Include tokenizer and action dimension parameters
- Maintain compatibility with existing Pi0 inference patterns

Technical Details:
- Extended Pi0 architecture for expert-enhanced multimodal reasoning
- Flow matching timestep injection with adaRMS normalization
- Vision-language-action model with discrete state integration

Code Quality:
- Applied Black code formatting
- Fixed trailing whitespace and line endings
- Ensured isort import organization compliance

Author: Teeeio

examples/pi0_5/conf/inference.yaml

@@ -0,0 +1,26 @@
+defaults:
+  - _self_
+  - inference: pi0_5
+
+experiment:
+  exp_name: PI05_Inference
+  seed: 42
+  exp_dir: outputs/${experiment.exp_name}
+  ckpt_format: torch
+  task:
+    type: inference
+    backend: pi0_5
+    entrypoint: flagscale/inference/inference_engine.py
+  runner:
+    per_node_task: false
+    no_shared_fs: false
+    rdzv_backend: static
+    hostfile: null
+  cmds:
+    before_start: echo "Starting PI05 Inference"
+  envs:
+    LOGLEVEL: "INFO"
+    CUDA_VISIBLE_DEVICES: "0"
+    OTEL_SDK_DISABLED: true
+
+action: run

examples/pi0_5/conf/inference/pi0_5.yaml

@@ -0,0 +1,37 @@
+engine:
+  device: cuda
+  loader: pi0_5
+  model: /share/pi0_5/pi05_base
+  output_format: image
+  results_path: ${experiment.exp_dir}/results
+  stat_path: /share/pi05_dataset/stats.json
+  tokenizer: /share/paligemma-3b-pt-224
+  torch_dtype: float32
+  transformations:
+    LogIOTransformation:
+      log_level: info
+    StateScopeTransformation: {}
+experiment:
+  exp_dir: outputs/${experiment.exp_name}
+  exp_name: pi0_5_inference
+  seed: 42
+generate:
+  action_dim: 32
+  action_steps: 16
+  batch_size: 2
+  discrete_state_input: true
+  images_keys:
+  - observation.images.camera0
+  - observation.images.camera1
+  - observation.images.camera2
+  images_shape:
+  - 3
+  - 480
+  - 640
+  instruction:
+    task:
+    - retro serie of different cars with different colors and shapes, mdjrny-v4 style
+    - retro serie of different cars with different colors and shapes
+  tokenizer_max_length: 200
+  pi05_mode: true
+  state_key: observation.state