Improve simplified process tree

[edan-bainglass]

This PR does the following:

• Refactor tree classes to aiidalab_qe/common/process module
• Add missing "not rendered" guards for monitor-triggered events
• Improve tree styles, particularly height (max 500px)
• Improve performance of counting finished jobs by replacing recursion with single DB query
• Use a flag to signal when adding branches to avoid race conditions with monitoring thread

Future plan

• Replace more iterative/recursive tree components with single DB query
• Fetch more info from plugins to avoid on-the-fly processing

[edan-bainglass]

@mikibonacci @superstar54 please test for duplicate nodes. I added a test, but not sure how reliable it is.

[superstar54]

Hi @edan-bainglass , thanks for the work. The test seems not reliable. I think the test needs to mock the _add_branches_recursive with some time duration (~ 5 seconds), then you start another thread to trigger the _add_branches_recursive simultaneously. The time duration is important to test whether the lock indeed works.. If this is not easy to implement, we can skip the test for the moment.

[edan-bainglass]

Hi @edan-bainglass , thanks for the work. The test seems not reliable. I think the test needs to mock the _add_branches_recursive with some time duration (~ 5 seconds), then you start another thread to trigger the _add_branches_recursive simultaneously. The time duration is important to test whether the lock indeed works.. If this is not easy to implement, we can skip the test for the moment.

I'll give it a try. I'm more concerned in the meantime not being able to resolve the compatibility issue with v2.3

[edan-bainglass]

@superstar54 @mikibonacci test improved and indeed shows that the flag is necessary and effective.

[superstar54]

The code starts the threading first, then click in main. Thus, it can not guarantee that the main will run first.

[edan-bainglass]

The monitor thread cannot add new branches to collapsed branches. Only after the main branch expands a parent branch and begins adding sub-branches does the monitor thread attempt to add sub-branches.

This was the root of the problem. Duplications happened when a monitor event happened shortly (and while) the user was expanding a parent node.

[superstar54]

LGTM