[Bug]: Tied weights lose Parameter identity after sharding and meta materialization

Checklist

• 1. I have searched the existing issues (https://gitcode.com/mindspore/hyper-parallel/issues)
• 2. I have read the relevant documentation.
• 3. I have created a minimal reproduction case that clearly demonstrates the issue, including a complete code example and the error message with full traceback and error logs.

🐛 Describe the bug

问题描述

模型中的 tied weights 在参数分片或 meta device 物化后，可能只共享底层
storage，但不再是同一个 Parameter 对象。

例如 embedding 和 lm_head 完成分片后，原有逻辑只替换底层 tensor：

tensor_a = param_a.to_local() if isinstance(param_a, DTensor) else param_a.data

if isinstance(param_b, DTensor):
    param_b._local_tensor = tensor_a
else:
    param_b.data = tensor_a

此时虽然两端可能共享数据，但仍然存在两个独立的参数对象：

model.embed_tokens.weight is not model.lm_head.weight

这会影响 tied weights 的梯度聚合、优化器状态复用以及 HSDP 对共享参数
所有权的识别。

另外，meta 模型通过 to_empty 物化后，没有重新恢复 tied weights：

model.to_empty(device=device)
return model

to_empty 可能为同一绑定参数的不同 FQN 创建独立对象，导致 checkpoint
加载前的参数发现看到错误的参数关系。

期望行为

Expected behavior

参数分片及 meta materialization 完成后：

model.embed_tokens.weight is model.lm_head.weight

应继续成立。

对于 HSDP 模型，需要在 checkpoint target discovery 之前刷新 HSDP
参数状态并恢复共享参数；同时应校验绑定参数的 shape、dtype、DeviceMesh和 placements 是否一致，避免错误布局被静默绑定。

Additional context

Environment info

atlas a2/, torch 2.9.0

Thanks for contributing 🎉!

schema_version: 1
source: gitcode
gitcode_repo: mindspore/hyper-parallel
gitcode_issue: 388
source_url: https://gitcode.com/mindspore/hyper-parallel/issues/388

Trace the parameter sharding path shown in the issue and the meta-model to_empty materialization path first. Reproduce the embedding/lm_head case, then inspect checkpoint target discovery and HSDP parameter-state handling. Done means tied names reference the same Parameter after sharding and materialization, with incompatible shape, dtype, DeviceMesh, or placements rejected.