Set models to start in train mode by default

[bruAristimunha]

Ensure all models initialize in train mode instead of eval mode to facilitate training from the start.

closes: #744

[bruAristimunha]

I am changing some behaviour that you both defined a long time way for three nn.Sequential models, can you review if it makes sense for you guys, @robintibor and @gemeinl. The behaviour always starts the model with the evaluation model. This behaviour is always overridden if we are using skorch/eegclassifier, but this generates a silent error in PyTorch Lightning.

I will merge on Monday at the end of the day, as I am judging that the request made by @dungscout96 makes sense and the tests are green.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 87.75%. Comparing base (32d44b7) to head (56ea2b4).
Report is 1 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #745   +/-   ##
=======================================
  Coverage   87.75%   87.75%           
=======================================
  Files          94       94           
  Lines        7576     7577    +1     
=======================================
+ Hits         6648     6649    +1     
  Misses        928      928           

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.