Add Inception_v4 model config in Fluid API

[kuke]

Resolve #914

Ref. to Inception_v4 model in V2 API

[kuke]

Execute training on flowers data set (4 GPUs):

----------  Configuration Arguments -----------
batch_size: 128
init_model: None
lr_strategy: cosine_decay
model: inception_v4
num_layers: 50
parallel_exe: True
pretrained_model: None
with_mem_opt: True
------------------------------------------------
Pass 0, trainbatch 0, loss 4.6265411377,                        acc1 0.0078125, acc5 0.0234375 time 1.63 sec
Pass 0, trainbatch 10, loss 4.63355541229,                        acc1 0.0078125, acc5 0.0234375 time 1.79 sec
Pass 0, trainbatch 20, loss 4.61045503616,                        acc1 0.03125, acc5 0.0625 time 1.71 sec
Pass 0, trainbatch 30, loss 4.64154863358,                        acc1 0.0, acc5 0.015625 time 1.73 sec
Pass 0, trainbatch 40, loss 4.64159154892,                        acc1 0.0, acc5 0.0234375 time 1.66 sec
End pass 0, train_loss 4.61967611313, train_acc1 0.0224808678031, train_acc5 0.0508609712124,                test_loss 4.62793588638, test_acc1 0.0137033769861, test_acc5 0.0539314523339
Pass 1, trainbatch 0, loss 4.60240983963,                        acc1 0.0390625, acc5 0.046875 time 1.74 sec
Pass 1, trainbatch 10, loss 4.61039352417,                        acc1 0.03125, acc5 0.0546875 time 1.65 sec
Pass 1, trainbatch 20, loss 4.58687925339,                        acc1 0.0546875, acc5 0.0625 time 1.80 sec
Pass 1, trainbatch 30, loss 4.63385629654,                        acc1 0.0078125, acc5 0.03125 time 1.72 sec
Pass 1, trainbatch 40, loss 4.61811351776,                        acc1 0.0234375, acc5 0.0234375 time 1.70 sec
End pass 1, train_loss 4.61435079575, train_acc1 0.0272640306503, train_acc5 0.0497448965907,                test_loss 4.63180065155, test_acc1 0.00986013095826, test_acc5 0.0481665804982
Pass 2, trainbatch 0, loss 4.60251426697,                        acc1 0.0390625, acc5 0.0703125 time 1.74 sec
Pass 2, trainbatch 10, loss 4.58694887161,                        acc1 0.0546875, acc5 0.09375 time 1.69 sec
Pass 2, trainbatch 20, loss 4.61815309525,                        acc1 0.0234375, acc5 0.0390625 time 1.69 sec
Pass 2, trainbatch 30, loss 4.62593412399,                        acc1 0.015625, acc5 0.0234375 time 1.76 sec
Pass 2, trainbatch 40, loss 4.59237337112,                        acc1 0.046875, acc5 0.1015625 time 1.82 sec
End pass 2, train_loss 4.61462306976, train_acc1 0.0267857145518, train_acc5 0.046875,                test_loss 4.63678264618, test_acc1 0.0048828125, test_acc5 0.0441343262792
Pass 3, trainbatch 0, loss 4.61823987961,                        acc1 0.0234375, acc5 0.03125 time 1.71 sec
Pass 3, trainbatch 10, loss 4.61041259766,                        acc1 0.03125, acc5 0.046875 time 1.72 sec
Pass 3, trainbatch 20, loss 4.58694934845,                        acc1 0.0546875, acc5 0.0625 time 1.66 sec
Pass 3, trainbatch 30, loss 4.59472179413,                        acc1 0.046875, acc5 0.0859375 time 1.69 sec
Pass 3, trainbatch 40, loss 4.62600803375,                        acc1 0.015625, acc5 0.03125 time 1.82 sec
End pass 3, train_loss 4.611120224, train_acc1 0.0304528065026, train_acc5 0.0554846934974,                test_loss 4.63385057449, test_acc1 0.0078125, test_acc5 0.0470640137792