MAGN-9744 Studio 1.0.0.775 partially renders graphics on first try

[marimano]

Purpose

First of all, task name is not accurate because geometry is fully rendered after some time.
When there is a lot of geometry, Dynamo computes and renders it quite slowly. The PR adds some optimization to partly overcome the issue:

• SceneItems is constantly being called, so each time it generates new list and sorts it.
  In this PR SceneItems is computed only after changes in its source;
• To implement label behavior it had stored graphic array items separately to have an ability to find an array item by index (geometry doesn't know which node array item it belongs to) and such solution had slowed down rendering.
  In this PR node array geometry items are stored together as it had used to be and instead label positions are stored in a separate place, so when a node requests displaying label for its array item we know where to place the label.

Analysis

I made a competition between Dynamo without my changes and Dynamo with them. It was used lotsofstuff.dyn from the task.

• when code block value is 0..10..2:
  
  Render time for each node means time to generate geometry objects from render packages for the node.
  setting properties and rendering time means time to render already generated geometry. Render packages for each node are coming separately and between receiving packages, Helix renders what it already has.
  sum time means time passed from receiving the very first render packages to finishing render of all geometry.
  So, it this case we have 9 seconds of gain in time.
• when code block value is 0..20..2:
  
  In this case designations are the same, code block just gives much more array items, we have 1 minute and 19 seconds of gain in time.

These time values are got from output window in Visual Studio. Also, I noted the time on my smartphone starting at pressing Run button and stopping at all geometry appears. The results are next:

• when code block value is 0..15..2, not optimized Dynamo takes 1 minute and 45 seconds and optimized Dynamo takes only 40 seconds, so gain in time is 1 minute and 5 seconds;
• when code block value is 0..20..2, not optimized Dynamo takes 3 minutes and 20 seconds and optimized Dynamo takes 1 minute and 44 seconds, so gain in time is 1 minute and 36 seconds;

By the way computing render packages for a node with many array items is time-consuming operation as well. The PR does not consider this component of the delay in rendering

Declarations

• The code base is in a better state after this PR
• Is documented according to the standards
• The level of testing this PR includes is appropriate
• User facing strings, if any, are extracted into *.resx files
• All tests pass using the self-service CI.
• Snapshot of UI changes, if any - nothing is changed in UI

Reviewers

@mjkkirschner

FYIs

@jnealb @kronz @ikeough

[marimano]

For some reasons Camera is null in tests when running in self service, after adding this null check tests passed successfully:

[ramramps]

LGTM

[ikeough]

@marimano Are we just fixing the test to match the behavior here, or have the changes you made actually result in there being less objects in the scene?

[marimano]

@ikeough these changes in tests are done to match new behavior which is presented in this PR. All points of a node are stored in a single geomentry instead of storing each array item of the node separately. So, here are 6 points which are stored in 1 PointGeometryModel3D from now

[ikeough]

@marimano Thanks very much for this fix. The numbers look good! I had one question about the tests.

[jnealb]

@marimano BTW: where you say "Competition" we usually refer to that process as "Benchmark"