feat: lowering Triton Call

[avik-pal]

No description provided.

[github-actions]

EnzymeJAX Benchmarks

Benchmark suite	Current: defe9ed	Previous: 8a0664a	Ratio
scatter_sum / JaX / cpu / Primal	0.000004410500999074429 s	0.0000045017170021310446 s	0.98
scatter_sum / JaXPipe / cpu / Primal	0.000004263196099782362 s	0.000004457445099251344 s	0.96
scatter_sum / JaX / tpu / Primal	0.0001518684996 s	0.0001498850076983 s	1.01
scatter_sum / JaXPipe / tpu / Primal	0.0001430466716003 s	0.0001502703774021 s	0.95

This comment was automatically generated by workflow using github-action-benchmark.

[avik-pal]

enzymexlamlir-opt: external/llvm-project/llvm/lib/MC/MCSubtargetInfo.cpp:63: void ApplyFeatureFlag(FeatureBitset &, StringRef, ArrayRef<SubtargetFeatureKV>): Assertion `SubtargetFeatures::hasFlag(Feature) && "Feature flags should start with '+' or '-'"' failed.                                                                                                                                                                                                                        
PLEASE submit a bug report to https://github.com/llvm/llvm-project/issues/ and include the crash backtrace and instructions to reproduce the bug.                                                                                             
Stack dump:                                                                                                                                                    
0.      Program arguments: ./bazel-bin/enzymexlamlir-opt --triton-augment-function-with-extra-arguments --lower-triton --lower-kernel --canonicalize --lower-triton-extension-ops --mlir-print-ir-after-all --mark-func-memory-effects --canonicalize "--lower-jit=cuOptLevel=2 cubinFormat=bin indexBitWidth=32  cubinChip=sm_120 cubinFeatures=86 run_init=true toolkitPath=/home/avik-pal/.julia/artifacts/6d7c82e8a388df99ca324e4aa06700cb4464b489/lib/cuda" --symbol-dce test/lit_tests/triton/lowering.mlir                                                                                                                                                                                                                                                                                                                                                                                                                                                                          
Stack dump without symbol names (ensure you have llvm-symbolizer in your PATH or set the environment var `LLVM_SYMBOLIZER_PATH` to point to it):