[Dev] Add support and test case for Ladder Weight only Transformation Matmul Operator

[LeiWang1999]

This pull request includes several changes to the bitblas library, focusing on improving the matrix multiplication operations and adding new scheduling capabilities. The most important changes involve updates to propagation handling, scheduler conditions, and test configurations.

Propagation Handling:

• bitblas/ops/general_matmul/__init__.py: Updated the propagation handling to use TransformKind.LDMatrixTransform for boolean propagation and added a TODO comment to check device compatibility for propagation. [1] [2]

Scheduler Conditions:

• bitblas/ops/general_matmul/tilelang/dense/__init__.py: Split conditions for can_apply_fine_grain_scheduler and added a new can_apply_weight_propagation_scheduler function to handle specific propagation scenarios. [1] [2]
• bitblas/ops/general_matmul/tilelang/dense/__init__.py: Updated error message in the scheduler to provide more context on unsupported configurations.

Scheduler Class:

• bitblas/ops/general_matmul/tilelang/dense/matmul_tensorcore.py: Refactored MatmulWeightPropagationScheduler to inherit from MatmulFineGrainScheduler and removed redundant configuration methods.

Typing and Method Signatures:

• bitblas/tl/base_hint.py: Added type hints and converted from_roller_hint to a class method. [1] [2]

Test Configurations:

• testing/python/operators/test_general_matmul_ops_backend_tl.py: Updated the matmul_finetune function and tests to include the propagate_b parameter for more flexible testing. [1] [2] [3]