chore(client): basic query testing using mock net

[rvagg]

I've been working on some deeper testing as I iterate toward unixfs pathing (I really want to be able to do graph transfers to test that we're getting exactly the blocks we want). This is some of my initial forays into network-level testing.

• I realised that we never gave Query or QueryResult the bindnode treatment so I did that in here but expect that to be extracted out to a go-markets-types repo in the near future; in the process I replaced the "rpc" stuff with some go-ipld-prime processing (using feat(dagcbor): mode to allow parsing undelimited streamed objects ipld/go-ipld-prime#490, hence the untagged dependency).
• During testing I became concerned about messaging errors not properly resulting in a quick error turnaround; so you'll see some goroutine work in the client query to try and deal with that. It also tries to work in some additional context cancellation handling too because we use that a lot, although I hope that's largely dealt with by libp2p already. Maybe I've gone overboard here?

[codecov-commenter]

Codecov Report

Merging #83 (7d3689c) into main (1bab381) will increase coverage by 3.94%.
The diff coverage is 77.96%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #83      +/-   ##
==========================================
+ Coverage   44.34%   48.29%   +3.94%     
==========================================
  Files          28       29       +1     
  Lines        2546     2578      +32     
==========================================
+ Hits         1129     1245     +116     
+ Misses       1371     1274      -97     
- Partials       46       59      +13     

Impacted Files	Coverage Δ
pkg/client/client.go	25.88% <69.44%> (+25.88%)	⬆️
pkg/types/query.go	91.30% <91.30%> (ø)

[hannahhoward]

I would also argue that anything that runs on a mock net should live in an itest directory outside the hierarchy.

We should be ideally making the client have more abstract deps so it can be used independently.

I'm going to go ahead and merge cause it's certainly an improvement, but good feedback as you continue to make progress.

[hannahhoward]

the complexity of this simple response points to a challenging long term issue when we build these -- I don't see how we respond to actual graphsync requests with filecoin retrieval negotiation happening on top without putting actual graphsync and the go-fil-markets retrieval server on top. Which don't get me wrong -- we can do, I just want to identify how hard that might end up being.

[hannahhoward]

I would also argue that anything that runs on a mock net should live in an itest directory outside the hierarchy. The solution for a test that lives in 'client' is to make it more injectable so we can unit test it.

I'm not religious about this though.