WandbLogger does not log test results

[chris-clem]

🐛 Bug

The WandbLogger does not log test results when testing happens right after training.

To Reproduce

When running the MNIST example with the WandbLogger like in the following snippet, the test results do not get logged to wandb because it syncs before testing starts:

Code sample

    ...
    trainer = pl.Trainer(logger=WandbLogger(...))
    trainer.fit(model)
    trainer.test()

Expected behavior

WandbLogger should also log the test results by default like TensorBoardLogger or TestTubeLogger to be a real alternative for the other loggers.

Environment

PyTorch version: 1.4.0
Is debug build: No
CUDA used to build PyTorch: 10.1

OS: Ubuntu 18.04.4 LTS
GCC version: (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0
CMake version: version 3.10.2

Python version: 3.7
Is CUDA available: No
CUDA runtime version: No CUDA
GPU models and configuration: No CUDA
Nvidia driver version: No CUDA
cuDNN version: No CUDA

Versions of relevant libraries:
[conda] pytorch-lightning         0.7.1                    pypi_0    pypi
[conda] torch                     1.4.0                    pypi_0    pypi
[conda] torchvision               0.5.0                    pypi_0    pypi

[Borda]

@borisdayma @colehurwitz @calclavia pls ^^

[borisdayma]

@Borda I think it's due to the finalize method closing the run with wandb.join.

I'm not completely familiar with the typical use case but a possible solution would be to remove it from the callback.

[borisdayma]

@chris-clem can you confirm that this proposed fix solves the issue?

If so then related PR should fix it.

[Borda]

also discussed in #1116 and a fix in #1193
@borisdayma @chris-clem could you pls review it

[borisdayma]

The PR looks good to me!

[chris-clem]

The fix works, thanks!