chore: update news for scaffoldings

[garrett4wade]

No description provided.

[gemini-code-assist]

Code Review

This pull request adds a new news entry to the README regarding the integration with Scaffolding for agentic RL training and reorganizes the 'Previous Releases' section. Feedback was provided to maintain date format consistency, correct the framework's name, and improve readability by reducing word repetition.

[gemini-code-assist]

The new news entry has a few minor issues with consistency and phrasing:

1. Date Format: The date 2026-04-23 uses hyphens, whereas all other entries in the News section use slashes (e.g., 2026/04/18). It should be updated to 2026/04/23 for consistency.
2. Terminology: The framework is referred to as "Scaffolding" in the linked repository and the local directory. "Scaffoldings" (plural) is slightly inconsistent with the official naming.
3. Clarity: The word "enables" is used twice in consecutive sentences. Replacing one instance improves readability.

Suggested change

	**\[2026-04-23\]** 🚀 We’re excited to release our integration with
	[Scaffoldings](https://github.com/NVIDIA/TensorRT-LLM/tree/main/tensorrt_llm/scaffolding)
	for agentic RL training - now live in our
	[examples](https://github.com/inclusionAI/AReaL/tree/main/examples/scaffolding)! Huge
	shoutout to @narutolhy and @WeiHaocheng for making this happen 🙌. The modular design of
	the Scaffoldings enables it to achieve a thorough decoupling of agent execution, reward
	calculation, and trajectory acquisition. This enables developers to reuse existing
	modules when implementing an agentic RL method, allowing them to focus on their own
	innovative modules.
	**\[2026/04/23\]** 🚀 We’re excited to release our integration with
	[Scaffolding](https://github.com/NVIDIA/TensorRT-LLM/tree/main/tensorrt_llm/scaffolding)
	for agentic RL training - now live in our
	[examples](https://github.com/inclusionAI/AReaL/tree/main/examples/scaffolding)! Huge
	shoutout to @narutolhy and @WeiHaocheng for making this happen 🙌. The modular design of
	Scaffolding enables a thorough decoupling of agent execution, reward
	calculation, and trajectory acquisition. This allows developers to reuse existing
	modules when implementing an agentic RL method, focusing on their own
	innovative modules.