Update the codecov.yml

Signed-off-by: Jingyu Xin <jingyux@nvidia.com>

Author: jingyu-ml

.github/codecov.yml

@@ -11,15 +11,3 @@ coverage:
         target: auto
         threshold: 1% # Allow atmost 1% coverage drop from main branch.
     patch: false
-
-# Exclude GPU-only Triton kernel files from ALL codecov calculations (project
-# and patch checks, all flags). Rationale: these files are dominated by
-# @triton.jit kernel bodies that CPU unit tests cannot exercise. GPU tests
-# cover them end-to-end (see tests/gpu/torch/sparsity/attention_sparsity/) but
-# the `gpu`-flag upload may race with the PR status check, so relying on flag
-# combination alone leaves the project check flaky. Dropping these files here
-# makes the check deterministic — local `pytest --cov` and GPU runs still
-# measure them; only the codecov PR status ignores them.
-ignore:
-  - "modelopt/torch/kernels/triton_fa.py"
-  - "modelopt/torch/kernels/hf_triton_attention.py"