Robust RecursiveFindContent | Continue the search if data transfer fails

[KolbyML]

Clients can have bugs, trin has had bugs, shisui has, ultralight etc. A bug in a client shouldn't stop the whole show. A bug we were seeing was trin found the content from a shisui or trin node, then the uTP transfer would fail for some unknown reason, then we would just return an error.

Instead of this we could continue the search until we find a working peer, we already did a big search to find the contents neighbourhood so it isn't out of question if the first node we try to get the data from fails, we just try another and continue the search to some failure level till we either fail X times and give up or successfully fetch the data.

[morph-dev]

@carver if I understood correctly, you said you are looking into this or something similar, right? Maybe you can add your progress or assign it to yourself if that's the case

[carver]

Tracking To-dos before this is complete:

• Continue the RecursiveFindContent query until the uTP transfer is complete and the content is validated #1395
• Continue searching for content when an earlier transfer of content fails, or is invalidated #1467

Probably move to a follow-up issue:

• increase parallel to 2 eh, maybe not worth the added complexity
• Add peer failures to query traces #1479
• Remove Clone derivation from Query #1482