在gcn中use_residual与aggregator的self_embedding是否重复，或者区别是啥

file1: encoders.py: 204 line
for hop in range(self.num_layers - layer):
if self.use_residual:
h = hidden[hop] + \
aggregator((hidden[hop], hidden[hop + 1], adjs[hop]))
else:
h = aggregator((hidden[hop], hidden[hop + 1], adjs[hop]))
next_hidden.append(h)
hidden = next_hidden

file2:sparse_aggregators.py:78 line
if self.concat:
return tf.concat([from_self, from_neighs], 1)
else:
return tf.add(from_self, from_neighs)

Start with encoders.py around line 204 and sparse_aggregators.py around line 78. Trace how use_residual and the aggregator's self-embedding path contribute to h, then compare their effects in both concat and add cases; done means the distinction or confirmed overlap is documented clearly.

Written by the indexing model from the issue text.