Your question about "project re-renders after reopening" is perfectly logical. Similar cache behaviors exist from (say) switching between proxy and original media.
There are separate render caches for proxy and original media, and if both are fully rendered, just switching between them should not cause parts of the cache to become invalid (IOW render dots should not appear). Yet it sometimes does.
There is nothing wrong with submitting a request to "Provide Final Cut Pro Feedback" on this topic, in fact I encourage this. You can get to that by using the menu Final Cut Pro > Provide Final Cut Pro Feedback...
Apple has said they read each one of these. Your particular request is a good use of the feedback tool, because it is self-contained, this issue is widely known, and IMO does not require an extensive replication scenario.
There are seemingly several separate manifestations of this, but you need not list all of those. Just list what you have observed and state the negative impact it has on your work.
Some examples of apparent anomalies in the FCP render cache system I have observed myself:
- Relaunching FCP can invalidate parts of the render cache
- Switching between proxy and original media mode can invalidate parts of the cache
- Certain third-party utilities, such as Neat Video, may appear fully cached, but the render cache is not used during export.
- Some built-in titles exhibit different caching behavior between the first render and subsequent renders. E.g., if Basic Titles are rendered to cache and FCP is relaunched, the timeline range will appear uncached (render dots appear). But if you then re-render that to cache and re-launch FCP a 2nd time, the render cacher persists. That is illogical.
From an FCP developer standpoint, they can argue that these nuances are not documented, are subject to change without notice between versions, hence are not viable candidates for a bug or improvement request.
However there is an opposing view (which I hold) that these are clearly illogical and indicate a loss of efficiency that negatively impacts editors. It is not necessary that every FCP behavior be documented before engineering takes corrective action. E.g, if Magnetic Mask exhibits exponential performance degradation on longer clips, improving that is a valid request. Just because the FCP manual doesn't make any promises about Magnetic Mask behavior doesn't mean an enhancement request that clearly shows a problem is invalid. It is appropriate. The case with FCP render cache durability is similar.
So I encourage you to give feedback on this using the "Provide Final Cut Pro Feedback" form.