feat: add Step-Audio 2 project with external link

- Add Step-Audio 2 project linking to StepFun official page
- Remove example projects (pandas, pytorch, scikit)
- Project highlights CoT reasoning, reinforcement learning, and SOTA benchmarks

Author: FlyTia

content/projects/pandas/featured.png

@@ -0,0 +1,22 @@
+---
+title: Step-Audio 2
+date: 2025-07-23
+external_link: https://www.stepfun.com/research/en/step-audio2
+tags:
+  - Audio LLM
+  - Speech Understanding
+  - Chain-of-Thought
+  - Reinforcement Learning
+summary: End-to-end multimodal large language model with audio reasoning capabilities, achieving SOTA performance on multiple benchmarks.
+---
+
+Step-Audio 2 is an end-to-end multimodal large language model engineered for industrial applications. This model innovatively integrates a latent space audio encoder with audio reinforcement learning.
+
+**Key Features:**
+- Genuine end-to-end architecture eliminating traditional ASR+LLM+TTS pipelines
+- CoT-reinforcement learning fusion enabling audio reasoning capabilities
+- Acoustic knowledge enhancement via web search and audio retrieval
+- SOTA performance on MMAU, URO-Bench, and multiple ASR benchmarks
+
+[View Full Project →](https://www.stepfun.com/research/en/step-audio2)
+