[PyTorch Debug] Support log fp8 tensor stats for blockwise recipe

[lengerfulluse]

Support log fp8 tensor stats for fp8 blockwise recipe.

fp8_recipe: 'blockwise'

Description

When enable fp8 underflow logging for blockwise recipe, it report the following exception:

[rank2]:   File "/TransformerEngine/transformer_engine/pytorch/module/layernorm_linear.py", line 283, in forward
[rank2]:     weightmat = module.get_weight_workspace(
[rank2]:   File "/TransformerEngine/transformer_engine/pytorch/module/base.py", line 1263, in get_weight_workspace
[rank2]:     out = quantizer.quantize(tensor, dtype=workspace_dtype)
[rank2]:   File "/TransformerEngine/transformer_engine/debug/pytorch/debug_quantization.py", line 332, in quantize
[rank2]:     self._call_inspect_tensor_api(tensor, rowwise_gemm_tensor, columnwise_gemm_tensor)
[rank2]:   File "/TransformerEngine/transformer_engine/debug/pytorch/debug_quantization.py", line 247, in _call_inspect_tensor_api
[rank2]:     debug_api.transformer_engine.inspect_tensor_postquantize(**args)
[rank2]:   File "nvidia-dlfw-inspect/nvdlfw_inspect/base.py", line 257, in route_api
[rank2]:     ret = getattr(self.namespace_features[feat_name], api_name)(
[rank2]:   File "/usr/local/lib/python3.10/dist-packages/transformer_engine/debug/features/log_fp8_tensor_stats.py", line 113, in inspect_tensor_postquantize
[rank2]:     assert type(tensor) in [Float8Tensor, Float8TensorBase, MXFP8Tensor, MXFP8TensorBase], (
[rank2]: AssertionError: [NVTORCH INSPECT ERROR] Tensor weight must be a quantized tensor when using log_fp8_tensor_stats. Use log_tensor_stats for high precision tensors.

Please include a brief summary of the changes, relevant motivation and context.

Fixes # (issue)
Import the necessary Blockwise Tensor and Base.

Type of change

• Documentation change (change only to the documentation, either a fix or a new content)
• [x ] Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Infra/Build change
• Code refactoring

Changes

Please list the changes introduced in this PR:

• Change A
• Change B

Checklist:

• I have read and followed the contributing guidelines
• The functionality is complete
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes