[Dev] Fix a bug in general matmul ops with zero

[LeiWang1999]

This pull request includes changes to the bitblas/ops/impl/matmul_dequantize_impl.py and testing/python/module/test_bitblas_linear.py files. The most important changes include the removal of the MatMulNTDequantizeEmitter class, refactoring the decode_func method, and modifying the testing functions in test_bitblas_linear.py.

Refactoring:

• bitblas/ops/impl/matmul_dequantize_impl.py: Removed the MatMulNTDequantizeEmitter class and its associated methods. This class was marked as needing refactoring and has been completely removed in this pull request.

Code simplification:

• bitblas/ops/impl/matmul_dequantize_impl.py: Simplified the decode_func method by removing the redundant assignment of E to last_output. The last_output is now directly computed without the intermediate E variable. [1] [2] [3]

Test modifications:

• testing/python/module/test_bitblas_linear.py: Refactored the testing functions to remove the use of pytest.mark.parametrize. The parameters are now directly passed to the testing functions. [1] [2] [3]

Bug fixes:

• testing/python/operators/test_general_matmul_ops.py: Fixed a bug in the matmul_torch_forward function where the matmul function was called with incorrect parameters. The matmul function now correctly uses all the permuted_inputs except the last one.