Missed Optimization: Failure to simplify (x*C1) - ((y+x)*C2) to x*(C1-C2) - (y*C2)

[Leo0506]

define i16 @src(i16 %arg0, i16 %arg1) {
  %v0 = add nsw i16 %arg1, %arg0
  %v1 = mul nsw i16 %arg0, 8697
  %v2 = mul nsw i16 %v0, 6436
  %v3 = sub nsw i16 %v1, %v2
  ret i16 %v3
}

define i16 @tgt(i16 %arg0, i16 %arg1) {
  %v0 = mul nsw i16 %arg0, 2261
  %v1 = mul nsw i16 %arg1, 6436
  %v2 = sub nsw i16 %v0, %v1
  ret i16 %v2
}

alive2: https://alive2.llvm.org/ce/z/HPYK6s
godbolt: https://godbolt.org/z/Ynr3MKooz
Pattern found in: https://github.com/dtcxzyw/llvm-opt-benchmark/blob/main/bench/ffmpeg/optimized/jrevdct.ll