This fixes the "stalling problem" where component list stops showing new components after some time, and never recovers.

How does this PR change the system?

This PR fixes a reactivity problem with our use of tanstack's setQueryData(). If a list has its cache busted (and gets requeried), tanstack no longer provides a reactive list at the top level (though each element appears reactive). So after this happens, our code that pushes new components into the list no longer triggers the UI to re-render. I'm not sure why tanstack does this, but they warn about this pattern here:

It might work at first but can lead to subtle bugs along the way.

Out of Scope:

• Removing the cache bust: We do not believe we should be busting the cache in the first place, as it's unperformant and unnecessary; but with this fix, we at least react correctly. Doing it at this level blocks the problem from all our list MVs that use the updateCache() pattern. We'll fix the performance issue another time.

How was it tested?

• Ran a script to create 250 components, load the workspace in 3 tabs, and obsessively reload them one by one. No stalls and everything shows correctly.
• Added an artifical cache bust locally to reliably repro the problem. It stopped reproing after this fix was applied.

In short: 🔗