Support for all KernelBench kernels

Nobody has claimed this yet.

This issue is only about conformance: lower correctly and generate output compatible (`allclose(0.01)`) with PyTorch eager. When using the `--torch-compile` option, there are fewer errors, so the list below is using that flag. The additional errors when using full export can be tracked elsewhere.

Torch-MLIR issues:
* [ ] https://github.com/llvm/torch-mlir/issues/4574: Level1 { 89 ~ 93 }
* [ ] https://github.com/llvm/torch-mlir/issues/4575: Level1 { 12, 96, 99 }, Level 2 { 52 }
* [ ] https://github.com/llvm/torch-mlir/issues/4578: Level 2 { 18, 22, 42, 43, 45, 51, 58, 64, 92 }
* [x] https://github.com/llvm/torch-mlir/issues/4604: Level 1 { 14 }
* [ ] https://github.com/llvm/torch-mlir/issues/4610 Level 1 { 10, 11 }

MLIR Issues:
* [ ] `LLVM ERROR: SmallVector unable to grow (9TB)`: Level 1 { 34 }, Level 2 { 17, 28, 79 }
* [ ] `RuntimeError: gather(): Expected dtype int32/64 for index`: Level 1 { 95 }
* [ ] Outputs don't match (expected `0.01`, got `1.00` ): Level 1 { 94, 98, 100 }

Look into level 3 and 4.

@adam-smnk @tkarna

Start with the listed Torch-MLIR and MLIR failures, using the --torch-compile path and the KernelBench levels named in the issue. Work through the remaining kernel errors and verify that lowered output is compatible with PyTorch eager using allclose(0.01); the issue is done when the listed kernels conform.

Written by the indexing model from the issue text.