[CI] Mark FA bwd as attn-bwd (#3261)

The results of flash attention backward pollutes the performance results
of flash attention forward.
![Screenshot 2025-01-24
182529](https://github.com/user-attachments/assets/2207ff8c-81ce-446e-9b8b-cd5c5d0d4700)
![Screenshot 2025-01-24
182720](https://github.com/user-attachments/assets/077863d3-ad63-4822-bfaa-be463a006bba)

Signed-off-by: Whitney Tsang <[email protected]>

Author: whitneywhtsang

.github/workflows/triton-benchmarks.yml

@@ -258,8 +258,8 @@ jobs:
           mv $REPORTS/attn-performance.csv $REPORTS/attn-bwd-performance.csv
 
           source ../../scripts/capture-hw-details.sh
-          python ../../scripts/build_report.py $REPORTS/attn-bwd-performance.csv $REPORTS/attn-bwd-triton-report.csv --benchmark attn --compiler triton --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL,MODE" --tflops_col Triton-TFlops --hbm_col "Triton-GB/s" --tag $TAG
-          python ../../scripts/build_report.py $REPORTS/attn-bwd-performance.csv $REPORTS/attn-bwd-xetla-report.csv --benchmark attn --compiler xetla --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL,MODE" --tflops_col XeTLA-TFlops --hbm_col "XeTLA-GB/s" --tag $TAG
+          python ../../scripts/build_report.py $REPORTS/attn-bwd-performance.csv $REPORTS/attn-bwd-triton-report.csv --benchmark attn-bwd --compiler triton --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL,MODE" --tflops_col Triton-TFlops --hbm_col "Triton-GB/s" --tag $TAG
+          python ../../scripts/build_report.py $REPORTS/attn-bwd-performance.csv $REPORTS/attn-bwd-xetla-report.csv --benchmark attn-bwd --compiler xetla --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL,MODE" --tflops_col XeTLA-TFlops --hbm_col "XeTLA-GB/s" --tag $TAG
 
       - name: Run Prefix Sums kernel benchmark
         if: ${{ steps.install.outcome == 'success' && !cancelled() && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'prefix_sums.py') }}