Add MobileNetV3 architecture for Detection

[datumbox]

Fixes #1999

Trained using the code committed at 11408d6.

The current temporary pre-trained model was trained:

python -m torch.distributed.launch --nproc_per_node=8 --use_env train.py\
    --dataset coco --model fasterrcnn_mobilenet_v3_large_fpn --epochs 26\
    --lr-steps 16 22 --aspect-ratio-group-factor 3

Submitted batch job 34945231

Accuracy metrics:

0: IoU metric: bbox
0:  Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.230
0:  Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.385
0:  Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.233
0:  Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.026
0:  Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.223
0:  Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.446
0:  Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.218
0:  Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.298
0:  Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.305
0:  Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.041
0:  Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.315
0:  Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.577

Validated with:

python -m torch.distributed.launch --nproc_per_node=8 --use_env train.py\
 --model fasterrcnn_mobilenet_v3_large_fpn --test-only --pretrained

Submitted batch job 34643680

Speed benchmark:
0.14 sec per image on CPU

[fmassa]

Looks great to me, thanks a lot Vasilis!