[Backend] cuda normalize and permute, cuda concat, optimized ppcls, ppdet & ppseg

[wang-xinyu]

PR types(PR类型)

Backend

Describe

• CUDA kernel for Color convert + Normalize + Permute
• Concat support GPU
• Applied this optimization to PPClas, PPDet and PPSeg
• Tested on PPLCNetV2, PPYOLOE and UNet.

[wang-xinyu]

PPClas End2end test on T4, TRT8.4 Latency in ms.

Version	Model	FP32	FP16	INT8
0.3.0	PP-LCNetv2	2.23	1.87	1.78
This	PP-LCNetv2	1.54	1.03	0.89

PPClas End2end improved ~30%.

[wang-xinyu]

PPDet End2end test on T4, TRT8.4 Latency in ms.

.	Model	FP32
Before	PPYOLOE	35.05
After	PPYOLOE	32.97

Inference is 32.6ms, so Preprocessing is 2.45ms -> 0.37ms, preprocessing is 6.6x faster.

[wang-xinyu]

PPSeg UNet PPInfer GPU backend, T4.

Latency in ms.

Input size	Preprocess latency, before	after
2048x1024	33.921	1.405
320x160	0.228	0.194

Preprocessing is 1.18x ~ 24.14x faster, depends on input size.