fix sample code diff for OP(retinanet_detection_output, retinanet_target_assign), fix `gamma` default value for OP(sigmoid_focal_loss) (#23744)

* fix sample code diff for OP(retinanet_detection_output, retinanet_target_assign), fix  default value for OP(sigmoid_focal_loss). test=develop

* fix float value. test=develop

* add int type for gamma and alpha.test=develop

* polish error message, test=develop

* fix if logic in dygraph, test=develop

* fix bug, test=develop

* enhance error msg. test=develop

* print invalid argment

* update comment, test=develop

* fix enforce not meet specification

* fix enforce not meet specification, test=develop

OP(retinanet_detection_output, retinanet_target_assign, sigmoid_focal_loss, deformable_roi_pooling) error message enhancement. test=develop (#23726)

* test=develop modify some op for dyg rcnn

* solve the conflict of ops with the same name. test=develop

This reverts commit abe3e690.
test=develop

* solve the conflict of ops with the same name. test=develop

* test=develop fix doc of detection's img_info param

Update iou_similarity op to support non-normalized bbox

* yolo_box OP add Attr(clip_bbox). test=develop

* Split some APIs from nn.py to loss.py

test=develop

* fix test_detection unit-test

test=develop

* refine 13 en doc

* Polish En doc for some APIs
* Update some comments and API.spec

Fix english doc api: sequence_slice, sigmoid_cross_entropy_with_logits, ssd_loss, DetectionMAP, detection_map (#20405)

* Fix english doc api: sequence_slice, sigmoid_cross_entropy_with_logits, ssd_loss, DetectionMAP, detection_map

* modify document of 11 APIs
test=develop
test=document_fix

* fix dtype to data type and description of name parameter

Fix english doc api, invloves the op of retinanet_target_assign, sigmoid_focal_loss and retinanet_detection_output. (#20250)

* test=document_fix
test=develop
Fix english doc api, invloves the op of retinanet_target_assign, sigmoid_focal_loss and retinanet_detection_output.

* test=document_fix
test=develop
remove Notice: this OP supports CPU mode only in english doc api of retinanet_target_assign and retinanet_detection_output

* test=document_fix
test=develop
fix API Difference for retinanet_target_assign and retinanet_detection_output

* fix API.spec conflicts in english doc api of retinanet_target_assign, sigmoid_focal_loss,  retinanet_detection_output
test=develop
test=document_fix

* fix API.spec conflicts in english doc api of retinanet_target_assign, sigmoid_focal_loss, retinanet_detection_output
test=develop
test=document_fix

* polish en api,test=develop,test=document_fix

* modified api.spec, test=develop,test=document_fix

* modified ops for arc,test=develop,test=document_fix

* update for 1.6, test=develop,test=document_fix

* update API.spec

* Update API.spec

test=develop,test=document_fix

* Update API.spec

test=develop,test=document_fix

* refine Normal Uniform documnet

* fix fluid.data code example, test=develop, test=document_preview, test=document_fix

* use None instead of -1 in shape, test=develop, test=document_preview, test=document_fix

* refine doc, test=document_fix

* add API.spec,test=develop,test=document_fix

* update API.spec

* merge_empty_lod_tensor, test=develop

* fix multiclass_nms, test=develop

* refine API.spec, test=develop

* add unittest case for fetch, test=develop

* add lod tensor test, test=develop

* return index for multiclass_nms, test=develop

* add api for multiclass_nms2

* update API.spc, test=develop

* refine api doc, test=develop

* fix test_detection.py, test=develop

* polish code, test=develop

* add more unittest case, test=develop

test=develop, test=document_preview

test=develop, test=document_preview

* fix transform matrix bug, test=develop

* modify API.spec

* modify roi_perspective_transform_op to output mask and transform matrix

* modify comment

* modify comment

* modify API.spec

* update API.spec

* remove no use header, test=develop

* resolve conflict