* fcos infer

* fcos training

* fix stgraph fcosloss

* fcos doc

* fix dy_fcos

* add comment

* train eval deploy fcos

* remove fcos config, add comment

* fix iter_tic
Co-authored-by: Nnemonameless <nemonameless@qq.com>
Co-authored-by: Ngeorgenemo <nifeng@pku.edu.cn>

* support mask_rcnn for kunlun

* minor

* update ppyolo_mobilenet_v3 large & small

* support visualdl in preprocess data

* support most preprocess op

* update solov2 some detail

* fix check paddle version

* check md5 before load weights.

* add enable_static()

* follow comments

* remove blank line

* support infer for SOLOv2

* add ppyolo mobilenetv3

* share train_batch_size from reader to model

* update floor_divide

* add makedirs for dumping config

* add solov2 model

* fix train batch size

* add solov2_r101_vd_fpn_3x

* fix batch size and update modelzoo

* refactor code of solov2

* fix deploy/python

* check md5 after weights download

* move train_batch_size sync from load_config to merge_config

* warning when YOLOv3Loss.batch_size setting

* add batch_size in YOLOv3Loss for backward compatable.

* sync TrainReader.batch to global train_batch_size

* fix voc mAP %.