Fix issue in converting Mixtral 8x7B checkpoints from HF to MCore and update doc

[yeahdongcn]

Testing Done

root@a9effa877dd0:/workspace/megatron-lm# python tools/checkpoint/convert.py --model-type GPT --loader mixtral_hf --saver mcore --target-tensor-parallel-size ${TARGET_TP_SIZE} --target-pipeline-parallel-size ${TARGET_PP_SIZE} --target-expert-parallel-size ${TARGET_EP_SIZE} --load-dir ${HF_FORMAT_DIR} --save-dir ${MEGATRON_FORMAT_DIR} --tokenizer-model ${TOKENIZER_MODEL}
...
received transformer layer 30
received transformer layer 31
received final norm
received output layer
 > padded vocab (size: 32000) with 0 dummy tokens (new size: 32000)
saving checkpoint at iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 in torch format
  successfully saved checkpoint from iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 [ t 1/1, p 4/4 ]
saving checkpoint at iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 in torch format
  successfully saved checkpoint from iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 [ t 1/1, p 4/4 ]
saving checkpoint at iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 in torch format
  successfully saved checkpoint from iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 [ t 1/1, p 4/4 ]
saving checkpoint at iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 in torch format
  successfully saved checkpoint from iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 [ t 1/1, p 4/4 ]
saving checkpoint at iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 in torch format
  successfully saved checkpoint from iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 [ t 1/1, p 4/4 ]
saving checkpoint at iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 in torch format
  successfully saved checkpoint from iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 [ t 1/1, p 4/4 ]
saving checkpoint at iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 in torch format
  successfully saved checkpoint from iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 [ t 1/1, p 4/4 ]
saving checkpoint at iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 in torch format
  successfully saved checkpoint from iteration       1 to /workspace/checkpoints/mixtral-mcore-TP1PP4EP8 [ t 1/1, p 4/4 ]
Done!
/usr/lib/python3.12/tempfile.py:1075: ResourceWarning: Implicitly cleaning up <TemporaryDirectory '/tmp/tmp7agu7eyr'>
  _warnings.warn(warn_message, ResourceWarning)