docs: make native Transformers the Nano adoption path

[LauraGPT]

Summary

• Make released Transformers 5.17.0 the first Nano Python path across four READMEs, installation, model selection, Model Zoo, tutorials, notebook links, Sphinx and website/Pages.
• Canonical examples and notebook: docs: make native Transformers the Nano starting point QwenAudio/Fun-ASR#171 (merged). Toolkit, -hf, -vllm, GGUF and 31-language MLT remain distinct.
• Refresh existing bilingual article and model entries; preserve the curated blog layout.

Verification

• Actual native checkpoint CPU float32: English URL, Chinese waveform, keywords and padded bilingual batch; CLI and notebook Python cells executed.
• Site: 481 passed in three partitions; final affected-site tests: 38 passed.
• Root contracts: 279 passed, 1 existing skip, 3 browser/model-dependent cases deselected.
• Sphinx links: 46 passed/no skips. Native browser desktop/mobile EN/ZH: 4 passed.
• 202 pages validated; repeated builds byte-identical.
• No GPU capacity, hosted Colab UI or public native Space inference claim.

Publication

Signed+DCO. Source/diff/hash/base backup saved before push. Native HF card published separately. Original HF card write is rejected by repo-specific token scope; main Space candidate is in complete local acceptance, not yet published.