Simplify `run.py` 🏃‍♀️

[KristinaUlicna]

PR contribution summary

Why is this PR useful / good for? Please describe the problem(s) you're trying to address.

• Allows node feature & edge property computation at runtime by:
  • Checking all graphs in provided train, valid and infer datasets
  • Checking if graph attribute storage is allowed in the config; error otherwise.
  • Computing & storing the features in the source graph if allowed.

List of proposed changes / linked issues & discussions

• [x] ✅ Resolves #297
• [x] ✅ Resolves #276 [duplicate]

What should a reviewer concentrate their feedback on?

• 🏃 Scripts to run - replicating the results with run.py
• 💻 Code quality
• 📝 Everything looks OK?

What type of PR is this? (check all applicable)

• 🪄 Feature
• 🐛 Bug fix
• 🧑‍💻 Code refactor / style
• 🔥 Performance Improvements

Added tests?

• 👍 yes, re-vitalised the commented tests + now expects test_run.pyto fail

Hints for the reviewer:

At review time, please try a bunch of files which do not have the node features & edge properties in them. Set the store_graph_attributes_permanently config hyperparameter to False 👇

store_graph_attributes_permanently: False

which should throw an error & instructions on what to do. Please follow the instructions until you get to the point where you can successfully train your model 😃

---

PR review summary

Describe what this PR does & how you reviewed the individual items, where needed:

Some helper checks to tick off:

• Focus on image annotation
• Focus on model training
• Could any optimization be applied?
• Is there any redundant code?
• Are there any spelling errors?

In conclusion, after my review, I'd like to:

• 🙋 ask some clarifying questions
• 🙅 suggest some specific changes

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[crangelsmith]

Question:

If I already have the graph attributes saved in my graph but I want to update them I should also set: store_graph_attributes_permanently: True?

In that case some suggestions:

• I think the config name should be a bit more clear that we are actually computing as well as storing these attributes.
• I think we should log clearly what is happening with the attributes in the console. e.g:
  • If we load a graph that already has attributes and the config is False, log that we are using precomputed attributes.
  • If we load a graph that already has attributes and the config is True, log that we are using recomputing the attributes that will be overwritten.
  • Similar to when you don't have attributes yet, the case where store_graph_attributes_permanently: False is very well documented/described but when is true we should log it too.

[KristinaUlicna]

After some offline discussions & design re-consideration, this PR lost on value as the issues can be approached differently, without needing to store the NODE_FEATURES vector in the written-out graph. This gets addressed in #329 . Closing this deprecated PR now.