bitnet: honor relu2 hidden_activation from GGUF (fixes microsoft/BitNet#602)

[msrovani]

Fixes microsoft/BitNet#602. The official bitnet-b1.58-2B-4T declares hidden_act: relu2 but the bitnet arch hardcodes LLM_FFN_SILU in build_ffn, producing wrong-but-finite logits on every backend.

Changes:

• llama-model.cpp: add "relu2" -> LLM_FFN_RELU_SQR to LLM_FFN_OP_TYPES_FROM_STRING
• models/bitnet.cpp: read .hidden_activation in load_arch_hparams (mirroring modern-bert), defaulting to LLM_FFN_SILU for backward compatibility with existing GGUF files; use hparams.llm_ffn_op in the graph

LLM_FFN_RELU_SQR with LLM_FFN_PAR gating produces exactly relu2(gate) * up, matching the HF BitnetMLP (act_fn(gate) * up).

[ggml-gh-bot]

Hi @msrovani, thanks for your contribution!

Per our contribution guidelines, the automated PR checker found the following issue(s) that need your attention:

• PR Template not respected: Please respect the template when creating a new pull request. Make sure to fill out all required sections.

---

Please note that maintainers reserve the right to make final decisions on PRs. If you believe there is a mistake, please comment below.