docs(README): restructure into one clean get-started flow (de-patchwork pass 2)

[LauraGPT]

Reorders sections so the README reads as one coherent flow instead of patchwork. Content unchanged except removing one duplicate example — sections are only reordered.\n\nNew order: Quick Start -> Installation -> Models -> Usage -> Deploy -> Benchmark -> What's new -> Community/....\n\nToday Benchmark and What's-new sit between Quick Start and Installation (breaking the get-started path) and Installation is buried at #5. Also removed the redundant explicit <a name> anchors (GitHub auto-anchors from the headings already provide #benchmark / #model-zoo) and the duplicate Fun-ASR-Nano 31 languages snippet in Usage.\n\nOpening for review, not merging (touches README structure).

[gemini-code-assist]

Code Review

This pull request restructures the README.md by moving the "Benchmark" and "What's new" sections to the bottom of the document and removing the <a name="model-zoo"></a> anchor. It also removes a code snippet demonstrating the usage of Fun-ASR-Nano-2512 with VAD configuration. The reviewer points out that removing this snippet leaves English users without an example of how to configure VAD for Fun-ASR-Nano, which is critical to prevent hallucinations and errors on audio files longer than 30 seconds. The reviewer suggests restoring this code snippet with an updated comment.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

I am having trouble creating individual review comments. Click here to see my feedback.

README.md (198-201)

Removing this snippet leaves the English README.md without any example of how to configure VAD for Fun-ASR-Nano.

Unlike SenseVoice or Paraformer, Fun-ASR-Nano is an LLM-based ASR model (Qwen3-0.6B decoder) and has a strict audio length limitation (typically 30 seconds). Running it on longer audio files without VAD chunking (specifically max_single_segment_time: 30000) will result in severe hallucinations, repetitions, or errors.

Since the Quick Start example for Fun-ASR-Nano in the English README (line 45) does not include vad_model="fsmn-vad", this Usage snippet is not a duplicate—it is the only place showing how to properly run Fun-ASR-Nano on longer audio files.

In contrast, the Chinese README (README_zh.md) retains both the VAD configuration in the Quick Start and this detailed Usage snippet. We should keep this example to ensure English users can successfully transcribe longer audio files.

# 31 languages with timestamps (requires VAD for audio longer than 30s)
model = AutoModel(model="FunAudioLLM/Fun-ASR-Nano-2512",
                  vad_model="fsmn-vad", vad_kwargs={"max_single_segment_time": 30000}, device="cuda")
result = model.generate(input="audio.wav", batch_size=1)