Fix eye dome lighting crashes

[lilleyse]

Fixes #9715 (local sandcastle)
Fixes #8818 (local sandcastle)
Fixes #6440 (local sandcastle)

Fixes several issues related issue to eye dome lighting using either destroyed or outdated shaders.

For #9715 the issue was basically:

• Attenuation was on so the derived command was created
• Attenuation was turned off (the derived command still exists)
• The style was changed. A new shader is set and the command becomes dirty. Problem now is that the derived command is still using the old shader.
• After the next frame the command stops being dirty
• Then attenuation is enabled again and it doesn't rederive the command because the command isn't dirty so it uses the old shader (either deleted or outdated)

The fix was to check if the derived command was using a different shader, and if so, rederive the command.

This is really just a symptom of the poor design of the derived command system which needs to be redone at some point.

CC @ebogo1 @mramato

[cesium-concierge]

Thanks for the pull request @lilleyse!

• ✔️ Signed CLA found.

Reviewers, don't forget to make sure that:

• Cesium Viewer works.
• Works in 2D/CV.
• Works (or fails gracefully) in IE11.

[ebogo1]

Thanks @lilleyse - I checked that these changes fix the crash and specs pass locally. Mainly just had a question but otherwise this looks good to me.

[ebogo1]

To make sure I understand this, does the else block only get reached on the first frame before pointCloudProcessor is given a value?

Also it might not help performance, but would it be more precise to wrap both the if and else here with an if (originalShaderProgram !== command.shaderProgram) so that the pointCloudProcessor object only gets updated when necessary? In Sandcastle doing this didn't seem to have any effect (as I'd expect) but maybe it's wrong if it's possible that the command or shader program need to get updated for a reason other than the mismatch we check for..

[lilleyse]

To make sure I understand this, does the else block only get reached on the first frame before pointCloudProcessor is given a value?

Yeah that's what's going on. I reorganized the code a bit so it is hopefully more self-documenting.

It should be ok to set originalShaderProgram every time even if the command is rederived for some other reason besides shader changes.

[ebogo1]

Thanks @lilleyse - just in case I made sure everything still works after the last commit but this looks good to me 👍