bugfix: Fix "more than one operator "/" matches these operands" (#1471)

<!-- .github/pull_request_template.md -->

## 📌 Description

Seeing the following error on GB200 on
https://github.com/flashinfer-ai/flashinfer/blob/fe442a2df64f46b021f3ad2bc184cd10b09b1d7d/include/flashinfer/attention/mla_hopper.cuh#L829

```
flashinfer/attention/mla_hopper.cuh(829): error: more than one operator "/" matches these operands:
            built-in operator "arithmetic / arithmetic"
            function "flashinfer::operator/(uint32_t, const flashinfer::uint_fastdiv &)" (declared at line 85 of flashinfer/attention/../fastdiv.cuh)
            operand types are: IdType / const flashinfer::uint_fastdiv
            (CAUSAL ? min(kv_end, kv_len - q_len + packed_qo_start / num_heads) : kv_end) /
                                                                   ^
          detected during:
            instantiation of "void flashinfer::mla::hopper::BatchMLAPageAttentionHopperKernel<KTraits,Params>(Params) [with KTraits=flashinfer::mla::hopper::HopperKernelTraits<true, 2U, 512U, 64U, 64U, 64U, DTypeQ, DTypeKV, DTypeO, IdType>, Params=flashinfer::MLAParams<DTypeQ, DTypeKV, DTypeO, IdType>]" at line 992
            instantiation of "cudaError_t flashinfer::mla::BatchMLAPageAttentionHopper<MASK_MODE,HEAD_DIM_CKV,HEAD_DIM_KPE,Params>(Params, uint32_t, uint32_t, cudaStream_t) [with MASK_MODE=flashinfer::MaskMode::kCausal, HEAD_DIM_CKV=512U, HEAD_DIM_KPE=64U, Params=flashinfer::MLAParams<DTypeQ, DTypeKV, DTypeO, IdType>]" at line 65 of fbcode/deeplearning/flashinfer/build/aot/generated/batch_mla_attention_dtype_q_f16_dtype_kv_f16_dtype_o_f16_dtype_idx_i32_head_dim_ckv_512_head_dim_kpe_64_profiler_False_sm90/batch_mla_sm90_run.cu
```

Adding `static_cast<uint32_t>()` to `packed_qo_start` resolve the issue
for me.



## 🔍 Related Issues

<!-- Link any related issues here -->

## 🚀 Pull Request Checklist

Thank you for contributing to FlashInfer! Before we review your pull
request, please make sure the following items are complete.

### ✅ Pre-commit Checks

- [ ] I have installed `pre-commit` by running `pip install pre-commit`
(or used your preferred method).
- [ ] I have installed the hooks with `pre-commit install`.
- [ ] I have run the hooks manually with `pre-commit run --all-files`
and fixed any reported issues.

> If you are unsure about how to set up `pre-commit`, see [the
pre-commit documentation](https://pre-commit.com/).

## 🧪 Tests

- [ ] Tests have been added or updated as needed.
- [ ] All tests are passing (`unittest`, etc.).

## Reviewer Notes

<!-- Optional: anything you'd like reviewers to focus on, concerns, etc.
-->

Author: 842974287

include/flashinfer/attention/mla_hopper.cuh

@@ -826,7 +826,8 @@ __global__ __launch_bounds__(KTraits::NUM_THREADS) void BatchMLAPageAttentionHop
           1 - (kv_start / CTA_TILE_KV);
 
       int mask_tile_idx =
-          (CAUSAL ? min(kv_end, kv_len - q_len + packed_qo_start / num_heads) : kv_end) /
+          (CAUSAL ? min(kv_end, kv_len - q_len + static_cast<uint32_t>(packed_qo_start) / num_heads)
+                  : kv_end) /
               CTA_TILE_KV -
           (kv_start / CTA_TILE_KV);