Fix model mapping issues and download resumption for HuggingFace models

[JingofXin]

📌 PR 内容 / PR Description

Fixed #839:

Wrong model mapping URLs like "stabilityai/stable-diffusion-3-medium"

Some Huggingface models require a token configuration, which can be done via: os.environ['LAZYLLM_MODEL_SOURCE_TOKEN'] = 'hf_ZBoxxxxxxx' (this needs to be done before importing lazyllm). If not configured, we've added a clearer error message

Wrong model names in official tutorial like "musicgen-small"

Fixed and added model mappings.

Builtin models are not working out of the box, like "internlm2-chat-7b-4bits"

Some examples are provided.

Cannot continue from interrupted model downloads

Fixed and removed the logic that would clear the data if the download was interrupted or failed.

🔍 相关 Issue / Related Issue

Fix #839

✅ 变更类型 / Type of Change

• 修复 Bug / Bug fix (non-breaking change that fixes an issue)
• 新功能 / New feature (non-breaking change that adds functionality)
• 重构 / Refactor (no functionality change, code structure optimized)
• 重大变更 / Breaking change (fix or feature that would cause existing functionality to change)
• 文档更新 / Documentation update (changes to docs only)
• 性能优化 / Performance optimization

🧪 如何测试 / How Has This Been Tested?

import os
os.environ['HF_ENDPOINT']="https://hf-mirror.com"
os.environ['LAZYLLM_MODEL_SOURCE'] = 'huggingface'
os.environ['LAZYLLM_MODEL_SOURCE_TOKEN'] = 'hf_ZBoOcccccccccxxxxxxxxxx'

import lazyllm
base = lazyllm.TrainableModule('stabilityai/stable-diffusion-3-medium')

Demo Use: internlm2-chat-7b-4bits
(1) vLLM

import lazyllm
chat = lazyllm.TrainableModule('internlm2-chat-7b-4bits').deploy_method(
    (lazyllm.deploy.vllm, {'max_model_len': 8192}))
chat.start()
res = chat("写一个关于春天的诗歌。")
print(res)

(2) LMdeploy

import lazyllm
chat = lazyllm.TrainableModule('internlm2-chat-7b-4bits').deploy_method(lazyllm.deploy.LMDeploy)
chat.start()
res = chat("写一个关于春天的诗歌。", stop='<|im_end|>')
print(res)

📷 截图 / Demo (Optional)