Increase MAX_TENSOR_NUM

The original setting in transformer_engine/common/transformer_engine.cpp is:

const size_t MAX_TENSOR_NUM = 20 * 1024 * 1024 / sizeof(Tensor);

When I used TransformerEngine for LoRA fine-tuning Qwen3-Next, which has many experts (512+1), I encountered this error:

transformer_engine.cpp:249 in function Allocate: Cannot allocate a new NVTETensor. Maximum number of tensors reached: 81920. There is probably a memory leak in your application.

We may need to increase this hard-coded limit.

Start in transformer_engine/common/transformer_engine.cpp and inspect the MAX_TENSOR_NUM constant and the Allocate path that reports the limit. Reproduce or validate against the reported LoRA fine-tuning workload for Qwen3-Next, then confirm that the new limit avoids the maximum-tensor error without introducing a new allocation failure.