network: unify wsPeerCore use for HTTP and p2p HTTP transport

[algorandskiy]

Summary

wsPeerCore refactoring aiming to make HTTPPeer to use "get path and fire request` approach rather than analyzing peer/url type.

Before:
wsPeerCore has rootURL field that is used for logging, peer identification and HTTP connectivity. Its constructor also accepts a RoundTripper (rate limiting transport) to construct a default http.Client.
Adding p2p into the mix changed the latest part - HTTP connectivity is handled by pre-created p2p http transport (and client) making the rootURL to serve a dual purpose (peer id as a full p2p multiaddr string, and a peer type separation field in HTTP clients code)

After:
wsPeerCore still has rootURL for id and connectivity but the connectivity bit is handled by pre-constructed http.Client that uses rootURL to build up a full URL. This allowed to unify peers with http and p2p transport from client's viewpoint.

Test Plan

1. Added a unit test for the new wrapper HTTPPAddressBoundTransport type
2. Existing tests should pass.

[codecov]

Codecov Report

Attention: 12 lines in your changes are missing coverage. Please review.

Comparison is base (472f01b) 56.08% compared to head (f0955ce) 56.06%.

Files	Patch %	Lines
network/wsNetwork.go	85.29%	5 Missing ⚠️
network/hybridNetwork.go	0.00%	4 Missing ⚠️
network/p2pNetwork.go	66.66%	2 Missing ⚠️
network/netprio.go	0.00%	1 Missing ⚠️

Additional details and impacted files

@@               Coverage Diff               @@
##           feature/p2p    #5933      +/-   ##
===============================================
- Coverage        56.08%   56.06%   -0.02%     
===============================================
  Files              481      481              
  Lines            68072    68053      -19     
===============================================
- Hits             38180    38157      -23     
- Misses           27281    27290       +9     
+ Partials          2611     2606       -5     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[gmalouf]

In general, I think this is an improvement so am marking approved. It would be nice if there was a rich type for address that handled whether its for http or multiaddr at construction (making it a property of a struct, removing some checks, etc) but not a show stopper.

[gmalouf]

Is there any benefit to consider around using a stronger type than string here? I.E. enforcing formatting etc

[algorandskiy]

I added this in previous PR, HTTPPeer type was needed only to get its address to reconstruct p2p transport. both ws and p2p implementations fail on parsing so I think it is OK

[algorandskiy]

Also thought about wrapping address - looked a bit bigger change, maybe in subsequent PRs as needed.