sync_inference_weights corrupts the sampler for dense Qwen3 models with tied embeddings

Minimal repro (dense Qwen3-4B-Instruct-2507, tie_word_embeddings: true):

1. Launch the trainer in full-weight mode on the exact dense-Qwen3 contract (ce_mode: bi_fused, fa4, merge_qkv: false) and xorl-sglang with --rl-on-policy-target xorl --enable-fp32-lm-head on the same model.
2. Fresh sampler generates perfectly (<guess>CRANE</guess> for the Wordle prompt).
3. add_inference_endpoint + sync_weights_to_inference(sync_method='nccl_broadcast') from the untrained trainer — i.e. byte-identical weights. Sync reports success: Synced 399 params to 1 endpoint(s), 8,822,848,512 bytes.
4. The same prompt now yields deterministic garbage ('arn讹tereg游客ashingtonoins Prompt trứng…' — identical string across repeated syncs/restarts, so the sync deterministically writes wrong bytes).

Likely suspect: tied embeddings. Qwen3-4B has tie_word_embeddings: true (Qwen3-8B does not); a scrambled lm_head/embedding readout produces exactly this symptom (coherent-frequency random tokens from step one). Possibly interacting with the exact-contract trainer's wrapped/Class-B parameter representations at extraction time.

Impact: blocks the endpoint-driven Wordle zero-K3 recipe (xorl-client examples/wordle) on tied-embedding dense models via full-weight sync. Found while reproducing the train-inference-mismatch write-up's Wordle experiment (see #84 thread for the K₃ measurements that motivated the full-weight route).

Env: 2×H100, torch 2.11, branch qywu/value-head-critic (sync path identical to main).

Start with add_inference_endpoint and sync_weights_to_inference(sync_method='nccl_broadcast'), using the dense Qwen3-4B-Instruct-2507 contract and the Wordle reproduction described here. Trace parameter extraction and synchronization for tied embeddings, then verify that syncing byte-identical untrained weights preserves the sampler output and no longer produces deterministic garbage.