[CI] update typos config for CI pre-commit and fix some spells (#20919)

Signed-off-by: Peter Pan <Peter.Pan@daocloud.io>

tests/kernels/moe/modular_kernel_tools/common.py

@@ -416,7 +416,7 @@ class RankTensors:
         # We dequant and use that as hidden_states so the tests are stable.
         # quantizing and dequantizing yield slightly different results
         # depending on the hardware. Here we, quantize and dequantize
-        # first - so further quantize and dequantize will yeild the same
+        # first - so further quantize and dequantize will yield the same
         # values.
         if config.is_per_tensor_act_quant:
             a_q, a_scales = ops.scaled_fp8_quant(

tests/kernels/moe/test_deepgemm.py

@@ -95,7 +95,7 @@ def run_single_case(m, n, k, topk, num_experts, block_size):
     topk_weights, topk_ids = torch.topk(router_logits, k=topk, dim=-1)
     topk_weights = torch.nn.functional.softmax(topk_weights, dim=-1)
 
-    # triton referrence
+    # triton reference
     out_triton = fused_experts(
         hidden_states=tokens_bf16,
         w1=w1,