[PyTorch Debug] Fix issue with start_end_list logging feature

[paul-gibbons]

Description

Debug tool in TE allows user to set non-overlapping list of (start, end) pairs in which logging will occur at specified frequency. This MR address two bugs with start_end_list functionality.

1. start_end_list was directly being used List in stats_buffer, leading to unhashable type error at the first logging interval. List is converted to nested tuples in LogTensorStats and LogFP8TensorStats which resolves the issue.

TransformerEngine/transformer_engine/debug/features/utils/stats_buffer.py", line 193, in try_add_buffer
[rank0]:     if (layer_name, tensor_name, options) in self.buffers:
[rank0]:        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[rank0]: TypeError: unhashable type: 'list'

2. Add additional handling in stats_buffer to clean up scenario where next_iter=None when using start_end_list and we face type error when checking iteration >= next_iter in stats_buffer.py below.

"/pg/dev/code/TransformerEngine/transformer_engine/debug/features/utils/stats_buffer.py", line 224, in log_stats
[rank0]:     if not self._if_run_reduction():
[rank0]:            ^^^^^^^^^^^^^^^^^^^^^^^^
[rank0]:   File "/pg/dev/code/TransformerEngine/transformer_engine/debug/features/utils/stats_buffer.py", line 184, in _if_run_reduction
[rank0]:     if iteration >= next_iter:
[rank0]:        ^^^^^^^^^^^^^^^^^^^^^^
[rank0]: TypeError: '>=' not supported between instances of 'int' and 'NoneType'

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