Fix trtllm-gen attention illegal memory access (#2002)

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

## 📌 Description

This PR fixes illegal memory access of trtllm-gen attention kernels. It
changes the workspace buffer from `int_workspace_buffer` to
`float_workspace_buffer`. `int_workspace_buffer` is a fixed sized buffer
and not initialized to zero, which should not be used.

<!-- What does this PR do? Briefly describe the changes and why they’re
needed. -->

## 🔍 Related Issues

Issue #1928 

## 🚀 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

- [x] I have installed `pre-commit` by running `pip install pre-commit`
(or used your preferred method).
- [x] I have installed the hooks with `pre-commit install`.
- [x] 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

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

## Reviewer Notes

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


<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

* **Bug Fixes**
* Fixed memory allocation in the decode module to improve computation
accuracy and stability during text generation.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->

Author: Tom-Zheng

flashinfer/decode.py

@@ -1988,7 +1988,7 @@ def paged_run(
             q.contiguous(),  # NOTE(Siyuan): without contiguous, the result is incorrect
             paged_k_cache,
             paged_v_cache,
-            int_workspace_buffer,
+            float_workspace_buffer,
             block_tables,
             kv_lens_buffer,
             max_kv_len,