Memory leak when log_max_attention_logits is set

**Describe the bug**

Currently, there's a flag in mcore to log the maximum value of QK attention logits in tensorboard `--log-max-attention-logit`. This flag and `--qk-clip` are the only flags that trigger obtaining and storing the max QK logit from TE in [TEDotProductAttention](https://github.com/NVIDIA/Megatron-LM/blob/8de8238844bb7824d3e245efae89d7c8c4211bc7/megatron/core/extensions/transformer_engine.py#L1628)

If QKClip is enabled, [`clip_qk()`](https://github.com/NVIDIA/Megatron-LM/blob/main/megatron/core/optimizer/qk_clip.py) uses the QKClip implementation in [SelfAttention](https://github.com/NVIDIA/Megatron-LM/blob/8de8238844bb7824d3e245efae89d7c8c4211bc7/megatron/core/transformer/attention.py#L1693) or [MLASelfAttention](https://github.com/NVIDIA/Megatron-LM/blob/8de8238844bb7824d3e245efae89d7c8c4211bc7/megatron/core/transformer/multi_latent_attention.py#L1148) to clip the weights after which `core_attention.current_max_attn_logits` is reset.

When `--qk-clip` is not used, the attention logits are no longer reset because `self_attention.clip_qk()` is never called. Additionally, as `torch.max()` is called without `torch.no_grad()` or `.detach()` in TEDotProductAttention, every batch's max logits gets added to the computational graph, thus causing a memory leak (on RAM, not GPU) that grows on every DPA call to TE.

**Steps/Code to reproduce bug**

In NeMo framework container 26.04, the attached train.py and repro.txt (standard hydra config file; GitHub didn't let me upload it as YAML) can be used. Attached config is a slightly modified version of llama 3's pretraining recipe available on mbridge. Basically if model.log_max_attention_logit is True but model.qk_clip is False, this bug occurs.

```bash
export CUDA_DEVICE_MAX_CONNECTIONS=1
export HYDRA_FULL_ERROR=1
export NCCL_NVLS_ENABLE=0
export HF_TOKEN=
export MLFLOW_ENABLE_SYSTEM_METRICS_LOGGING=true
export MLFLOW_DISABLE_TELEMETRY=true

python -u "/workspace/scripts/train.py" \
--config-path="/workspace/scripts/config" \
--config-name="repro" \
1>> "/workspace/logs/stdout.log" 2>> "/workspace/logs/stderr.log"
```

**Expected behavior**

CPU memory should not keep constantly increasing during the pretraining process.

**Additional context**

Possible solution: Either detaching the tensor in TE/mcore from the autograd graph or setting the logits to None even for the case where qk-clip is not enabled.

System memory usage graph illustrating the issue:

Here,
- control = both log_max_attention_logit and qk_clip set to False
- qk_clip = both log_max_attention_logit and qk_clip set to True
- repro = log_max_attention_logit alone set to True (qk_clip = False)

Attached: [train.py](https://github.com/user-attachments/files/27514892/train.py), [repro.txt](https://github.com/user-attachments/files/27514919/repro.txt)