Handle multiple failures in sled-agent firewall rule requests

[plotnick]

When firewall rules are sent to multiple sleds, only a single error is currently handled (as of PR #1636). We should have some form of long-running, retry-style error handling in case something goes wrong on more than one sled.

Originally raised by @bnaecker in #1636 (comment)

[ryaeng]

@plotnick Do you have any examples? I'd like to give this a shot, but I'm not familiar with implementing this sort of thing. Thanks.

[plotnick]

I do not, unfortunately. @bnaecker made the original suggestion (see link to #1636); maybe he has some ideas? But I think in general this would be quite a lot of work to implement fully.

A simpler first step might be to simply combine (maybe just as ,-joined strings, to start) any errors from the results gathered at

omicron/nexus/src/app/vpc.rs

Line 432 in 2ea34d0

let results = join_all(sled_requests).await;

That would at least let us report multiple failures instead of just returning the first one, and so would be an improvement over what's currently there.