1. Use CudnnWorkspaceHandle in exhaustive search of conv_cudnn.
2. For Ops using CudnnWorkspaceHandle in exhaustive search, release their GPU memory after exhaustive search.

test=develop

* make_conv_cudnn_ws_size_configurable, test=develop

* change std::max to std::min
test=develop

test=develop

test=develop

* Revert "set constant for loss"

This reverts commit 167933f6.

* Revert "remove workspace_handle"
test=develop
This reverts commit b4aca8ed.

* remove workspace_handle
test=develop

* set constant for loss
test=develop

* Inception fusion operator.
* Support horizontal layer fusion in conv_fusion_op.
* Search conv algo strategy for variable-length input.
   search N times and cache the searched algos. For other input, choose the algo of input whose area is closest to this input.

test=develop

fix conv+elemenwise fuse bug.

* Refine conv_fusion_op for identity activation.
* Fix unit testing.
test=develop

test=develop

* Convolution fusion operator.
* Clean code
test=develop