* refactor rbox

* modify the code of save results

* fix some problem

* add .gitignore in dataset/dota

* fix test anno path

* refactor s2anet to support batch_size > 1

* fix problem of inference

* support batch_size > 1 for training

* fix empty results

* fix dota eval

* fix configs of s2anet_head

* modify s2anet_spine_1x to 73 mAP

* updata document, test=document_fix

* use 8 card for training as default, test=document_fix

* set lr for 4 card as default, and update

* s2anet update

* remove np anchor

* support --eval for s2anet

* add spine dataset and config yml for s2anet

* add doc for s2anet

* update doc

* fix typo, test=document_fix

* add some comments, and update import

* Add S2ANet model for Oriented Object Detection.