Ok, let me try again in more detail:

I'm basically saying I like the ideas along the lines of B the most. A is already there to an extent (only within the same datacenter), and C... Well. I'd rather SE put in the effort to resolve underlying causes rather than develop extensive workarounds.

But even though I like B, you mentioned not being sure how possible these things were, and I'm not either - because SE seems to struggle resolving issues that look simple from the outside, even with extensive diagnostic data from the community.

A further suggestion, if SE really cannot resolve 2002 in queue and otherwise make the process more tolerable and accessible, would be to change the sound of the error so it is not the same as the queue refreshing. One would hope this is simple enough...