Patch d member after adding to class

[frmdstryr]

I've been working on a port of atom more optimized for enaml-web. Due to how observation works in my port it requires that members be attached to a class before observers can be added. This is only change to enaml needed for my port to work. This makes it consistent with the other use of patch_d_member in DeclarativeMeta.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 66.06%. Comparing base (7895e4c) to head (560f9bc).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #622      +/-   ##
==========================================
- Coverage   66.07%   66.06%   -0.01%     
==========================================
  Files         270      270              
  Lines       26584    26584              
  Branches     3901     3901              
==========================================
- Hits        17565    17564       -1     
  Misses       7975     7975              
- Partials     1044     1045       +1     

🚀 New features to boost your workflow:

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

[MatthieuDartiailh]

Would you mind elaborating on what is specific to enaml-web workflow and the kind of optimization you implemented ?

[frmdstryr]

Thanks!

There are a lot more child nodes getting added/removed than in a typical Qt application. The optimization is just a memory/speed tradeoff that switches the observer pool to use hashmaps instead of vectors.