Update references to use the new Model Registration API

[datumbox]

Depends on #6364

This PR updates all the __dict__[model_name] occurrences within our code with a get_model() call.

To confirm the reference scripts still work we run the same experiments as #5618 and #6241. Minor variations on accuracy are expected because we use multiple GPUs and batch-sizes > 1. The important think to verify is that the models are initialized properly and provide similar accuracies.

torchrun --nproc_per_node=2 train.py --model efficientnet_b0 --test-only --weights EfficientNet_B0_Weights.IMAGENET1K_V1
Test:  Acc@1 77.684 Acc@5 93.536

python -u train_quantization.py --test-only --weights ResNet18_QuantizedWeights.IMAGENET1K_FBGEMM_V1 --model resnet18 --device cpu
Test:  Acc@1 69.494 Acc@5 88.882

torchrun --nproc_per_node=2 train.py --model retinanet_resnet50_fpn --test-only --weights RetinaNet_ResNet50_FPN_Weights.COCO_V1
IoU metric: bbox
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.363
 Average Precision  (AP) @[ IoU=0.50      | area=   all | maxDets=100 ] = 0.557
 Average Precision  (AP) @[ IoU=0.75      | area=   all | maxDets=100 ] = 0.382
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.193
 Average Precision  (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.400
 Average Precision  (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.490
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=  1 ] = 0.314
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets= 10 ] = 0.500
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=   all | maxDets=100 ] = 0.540
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = 0.340
 Average Recall     (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.581
 Average Recall     (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.696

torchrun --nproc_per_node=2 train.py --model lraspp_mobilenet_v3_large --test-only --weights LRASPP_MobileNet_V3_Large_Weights.COCO_WITH_VOC_LABELS_V1
global correct: 91.3
average row correct: ['94.6', '84.8', '68.5', '75.6', '58.3', '38.4', '78.8', '54.1', '90.1', '37.9', '80.6', '60.7', '79.5', '77.7', '81.9', '87.5', '51.5', '81.1', '60.8', '84.2', '54.3']
IoU: ['90.3', '68.5', '56.2', '57.7', '47.8', '33.0', '72.1', '46.1', '79.2', '31.3', '66.0', '35.5', '61.7', '63.1', '71.4', '76.7', '28.9', '62.7', '44.3', '69.9', '45.7']
mean IoU: 57.5

torchrun --nproc_per_node 1 --nnodes 1 train.py --val-dataset sintel --batch-size 1 --dataset-root $dataset_root --model raft_large --weights Raft_Large_Weights.C_T_SKHT_V2
Batch-processed 1040 / 1041 samples. Going to process the remaining samples individually, if any.
Sintel val clean epe: 0.6502	1px: 0.9255	3px: 0.9701	5px: 0.9805	per_image_epe: 0.6501	f1: 2.6801
Sintel val final Total time: 0:02:40
Batch-processed 1040 / 1041 samples. Going to process the remaining samples individually, if any.
Sintel val final epe: 1.0195	1px: 0.8898	3px: 0.9493	5px: 0.9660	per_image_epe: 1.0190	f1: 4.5747

torchrun --nproc_per_node=8 train.py \
    --batch-size=64 --test-only \
    --data-path="/datasets/clean_kinetics_400/" \
    --clip-len 16 --frame-rate 15 --clips-per-video 5 \
    --cache-dataset \
    --model mc3_18 --weights="MC3_18_Weights.DEFAULT" 
 * Clip Acc@1 52.820 Clip Acc@5 75.228
 * Video Acc@1 63.950 Video Acc@5 84.125

[jdsgomes]

LGTM!