Calculate the average time for benchmark.

Fix nan when api adapt to 2.0-beta

Add first-order-model to applications

Add StarGAN-v2 style FID calculation

Move some model to ppgan, Add sr model

Refine print log and add args

Refine application code

Add RealSR to applications

Fix save error

Adapt to api 2.0 again

Add DeepRemaster

add --style command line option to let user choose stargan or gan-compression style (by default gan-compression style will be used).
move `dygraph.guard()` declaration into fid module for two reason: 1. the inference model didn't work in dygraph mode, so we dynamically choose whether to use dygraph mode after style is determined. 2. easier to use for end user (no need to call fluid.dygraph.guard() explicitly)

Adapt 2.0 api

Add DeOldify