* Add documentation of cross-compiling for iOS.

* Correst the typo in documentation of cross-compiling for raspberry pi.

* Set ANDROID_API to 21 when it is specified < 21 for arm64-v8a in build_android.sh.

* Check the input and print the usage in MergeModel.cpp.

and change the size of test to make unit test faster
refine comment and log of mkldnnlayer

and add branch net comparing with cpu result

By shell command

```bash
sed -i 's#ifdef PADDLE_ONLY_CPU#ifndef PADDLE_WITH_GPU#g' `find ./paddle/ -name '*.h' -o -name '*.cc' -o -name '*.cpp' -o -name '*.c' -o -name '*.cu'`
sed -i 's#ifndef PADDLE_ONLY_CPU#ifdef PADDLE_WITH_GPU#g' `find ./paddle/ -name '*.h' -o -name '*.cc' -o -name '*.cpp' -o -name '*.c' -o -name '*.cu'`
```

* golang pserver optimizer config for user

* update

* update

* update

* update

* update by comments

* fix errors

* fix errors

* default learning rate, temperary fix

* update

* using etcd as fault tolerant training

* update

* workable version, ft not tested

* small fix

* update

* remove TODO

Use CHECK instead of PCHECK, because PCHECK is used for errno.

Fix #2567

1. v2.layer.parse_network does not correctly handle the generation output.
2. GatherAgentLayer does not correctly handle generation output when batch_size > 1
3. Fix CustomStackTrace for rnn group