[refactor] Move trtllm_fp8_kv_kernel to triton_ops directory

[harvenstar]

PR Description:

Move trtllm_fp8_kv_kernel.py to python/sglang/srt/layers/attention/triton_ops/ for
better code organization, as suggested in #14553.

@b8zhong @Fridge003

Updated all import references accordingly.

Testing

Tested with serving benchmark - no regression observed.

Command:
CUDA_LAUNCH_BLOCKING=1 python3 -m sglang.launch_server --model-path Qwen/Qwen3-235B-A22B-Instruct-2507-FP8 --trust-remote-code --tp 4 --attention-backend trtllm_mha --kv-cache-dtype fp8_e4m3

python3 -m sglang.bench_serving \
  --backend sglang-oai \
  --dataset-name random \
  --random-input-len 1024 \
  --random-output-len 1024 \
  --random-range-ratio 0.98 \
  --num-prompts 80 \
  --max-concurrency 16

Results:
============ Serving Benchmark Result ============
Backend:                                 sglang-oai
Traffic request rate:                    inf
Max request concurrency:                 16
Successful requests:                     80
Benchmark duration (s):                  79.99
Total input tokens:                      81050
Total input text tokens:                 81050
Total input vision tokens:               0
Total generated tokens:                  81085
Total generated tokens (retokenized):    10185
Request throughput (req/s):              1.00
Input token throughput (tok/s):          1013.21
Output token throughput (tok/s):         1013.65
Peak output token throughput (tok/s):    1200.00
Peak concurrent requests:                23
Total token throughput (tok/s):          2026.85
Concurrency:                             15.92
----------------End-to-End Latency----------------
Mean E2E Latency (ms):                   15922.70
Median E2E Latency (ms):                 16067.41
---------------Time to First Token----------------
Mean TTFT (ms):                          414.21
Median TTFT (ms):                        282.49
P99 TTFT (ms):                           839.00
-----Time per Output Token (excl. 1st token)------
Mean TPOT (ms):                          15.31
Median TPOT (ms):                        15.47
P99 TPOT (ms):                           16.62
---------------Inter-Token Latency----------------
Mean ITL (ms):                           15.32
Median ITL (ms):                         13.45
P95 ITL (ms):                            13.82
P99 ITL (ms):                            28.76
Max ITL (ms):                            454.27

[gemini-code-assist]

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!

[harvenstar]

/tag-and-rerun-ci

[harvenstar]

a small followup pr to #14553, file position adjustment, pls review. @Fridge003 Thanks!!

[Fridge003]

/tag-and-rerun-ci

[b8zhong]