Added condition for different schedulers

[Om-Doiphode]

I have updated the configure_optimizers in the deepforest/main.py file to account for the different learning rate schedulers and also made the necessary changes in the deepforest_config.yml file.

[bw4sz]

Great thanks for this PR.

couple pieces

1. With every pull request, you want to write a pytest that asserts the new functionality. Something like here
   

   DeepForest/tests/test_main.py

   Line 178 in 199d3b7

   def test_train_no_validation(m):

   
   but where you change the config, using config_args when creating the deepforest object (https://deepforest.readthedocs.io/en/latest/ConfigurationFile.html#passing-config-arguments-at-runtime-using-a-dict) to show that the changes don't have unexpected consequences.

2. This change would be breaking for anyone with a previous version of a config file. Copy your new config to where it gets distributed with the rest of the package. @henrykironde this one of those long term maintenance things we want to automate on push, we should have two spots to update config files. But for now, its fine.

deepforest/data/deepforest_config.yml

3. Its good to have some arg defaults in case users have old custom configs. I think the existing code should stay in the code base above the new code as a default until a major release and mark it as to be deprecated.

https://github.com/Om-Doiphode/DeepForest/blob/6e959757f9bfac578031234b72ca2f4c6097efd9/deepforest_config.yml#L37

[Om-Doiphode]

HI @bw4sz, I have included 7 test functions in the test_main.py file to account for the different schedulers, updated the deepforest_config.yml files in other folders, and updated the documentation for the scheduler. I have run all the tests, and they have passed successfully.

[bw4sz]

Great, instead of writing 7 separate tests. Its one test with multiple parameters. Using pytest.mark.parameterize.

For example:

DeepForest/tests/test_main.py

Line 159 in dba368b

@pytest.mark.parametrize("architecture",["retinanet","FasterRCNN"])

[bw4sz]

Looks good, thanks.

[bw4sz]

Nice job @Om-Doiphode