Support dumplicate tensor in the tuple unpack

[zheng-ningxin]

The input may be reused in the Tuple_constuct node, for example: ['input1', 'input1', 'input2'].
In this case node.inputs will be ['input1', 'input2'] and trigger the assert assert len(node.inputs) == len(list(last_cpp.inputs())), errmsg.
So we use the cpp_node to get the debugName of the inputs.

[mwsunshine]

hi, ningxin,

I am wondering when will this bug be fixed according to your schedule? Meanwhile, will you release any kind of notification once it is fixed?

[zheng-ningxin]

hi, ningxin,

I am wondering when will this bug be fixed according to your schedule? Meanwhile, will you release any kind of notification once it is fixed?

@mwsunshine I guess this PR will be merged in a week and I'll send you a message once merged.

[mwsunshine]

hi, ningxin,
I am wondering when will this bug be fixed according to your schedule? Meanwhile, will you release any kind of notification once it is fixed?

@mwsunshine I guess this PR will be merged in a week and I'll send you a message once merged.

COOL!
Waiting for your message! So What I should do is to upgrade my nni to the latest version and try to prune the model once again?

[mwsunshine]

hi, ningxin,
I am wondering when will this bug be fixed according to your schedule? Meanwhile, will you release any kind of notification once it is fixed?

@mwsunshine I guess this PR will be merged in a week and I'll send you a message once merged.

hi, ningxin,

I installed the latest nni via source code today, but the error remains. Is it fixed in the latest version? Or I still need to wait for a period of time.