feat: fix peer renewal, change Filter keep alive

[weboko]

Problem

Peer renewal doesn't work as intended.

Current behavior - it renews a peer to a one with which light node has a connection already.
Expected behavior - renews to a peer which we don't have a connection.

This PR also adds decrease for Filter keep alive queries to 30 seconds as was proposed in one of the threads in waku-org/specs#18, and some code improvements.

Solution

Fix the behavior and add some improvements.

Notes

Tested in the dogfooding app - waku-org/lab.waku.org#68

[weboko]

this was moved from @waku/utils

@waku-org/js-waku-developers let's be cautious of not placing non public utils in @waku/utils

non public utils are those that are needed only by @waku/* packages family and/or test package.

[github-actions]

size-limit report 📦

Path	Size	Loading time (3g)	Running time (snapdragon)	Total time
Waku node	183.4 KB (+0.01% 🔺)	3.7 s (+0.01% 🔺)	18.6 s (-14.78% 🔽)	22.3 s
Waku Simple Light Node	183.53 KB (+0.06% 🔺)	3.7 s (+0.06% 🔺)	24.8 s (-0.68% 🔽)	28.4 s
ECIES encryption	23.12 KB (0%)	463 ms (0%)	10.5 s (+210.21% 🔺)	11 s
Symmetric encryption	22.58 KB (0%)	452 ms (0%)	4.7 s (-2.25% 🔽)	5.2 s
DNS discovery	72.49 KB (0%)	1.5 s (0%)	8.7 s (-51.92% 🔽)	10.2 s
Peer Exchange discovery	74.15 KB (0%)	1.5 s (0%)	17.4 s (+93.17% 🔺)	18.9 s
Local Peer Cache Discovery	67.68 KB (0%)	1.4 s (0%)	18 s (+1.39% 🔺)	19.4 s
Privacy preserving protocols	38.87 KB (0%)	778 ms (0%)	10.7 s (+20.76% 🔺)	11.5 s
Waku Filter	112.89 KB (+0.04% 🔺)	2.3 s (+0.04% 🔺)	23.3 s (+42.18% 🔺)	25.5 s
Waku LightPush	111.2 KB (-0.1% 🔽)	2.3 s (-0.1% 🔽)	20.8 s (+12.88% 🔺)	23 s
History retrieval protocols	111.73 KB (+0.11% 🔺)	2.3 s (+0.11% 🔺)	14.4 s (-26.06% 🔽)	16.6 s
Deterministic Message Hashing	7.29 KB (0%)	146 ms (0%)	3 s (+133.23% 🔺)	3.2 s

[danisharora099]

I'm curious how the peer renewal failure was observed/debugged? Any failing test cases? Should we add new tests that would reproduce it?

[weboko]

I'm curious how the peer renewal failure was observed/debugged? Any failing test cases? Should we add new tests that would reproduce it?

Peer renewal was working by dropping connection and then taking a peer from those with whom we have a connection. What we actually need to do is to get a peer to which we are not connected yet. Otherwise we need to always have connections to peers that we are not using in order to be able to use them as a replacement.

As for repro - you can do this by running dogfooding app now and seeing that at dome point there are no peers to be used for light push