chore: change amount of used peers to 2

[weboko]

Problem

As per RFC protocols should use 2 peers https://github.com/waku-org/specs/blob/master/standards/application/req-res-reliability.md#sending-with-redundancy

But 3 are used

Solution

Make them use 2

[github-actions]

size-limit report 📦

Path	Size	Loading time (3g)	Running time (snapdragon)	Total time
Waku node	86.47 KB (-0.05% 🔽)	1.8 s (-0.05% 🔽)	5.1 s (-33.98% 🔽)	6.8 s
Waku Simple Light Node	154.09 KB (-0.07% 🔽)	3.1 s (-0.07% 🔽)	5.6 s (-15.24% 🔽)	8.7 s
ECIES encryption	22.83 KB (0%)	457 ms (0%)	3.5 s (+162.53% 🔺)	3.9 s
Symmetric encryption	22.33 KB (0%)	447 ms (0%)	2.7 s (+10.22% 🔺)	3.2 s
DNS discovery	72.27 KB (0%)	1.5 s (0%)	5.3 s (+29.51% 🔺)	6.8 s
Peer Exchange discovery	73.75 KB (0%)	1.5 s (0%)	4 s (+33.24% 🔺)	5.5 s
Local Peer Cache Discovery	67.57 KB (0%)	1.4 s (0%)	3.5 s (-41.39% 🔽)	4.9 s
Privacy preserving protocols	38.91 KB (0%)	779 ms (0%)	3.8 s (+43.18% 🔺)	4.6 s
Waku Filter	81.07 KB (+0.01% 🔺)	1.7 s (+0.01% 🔺)	5.6 s (+41.91% 🔺)	7.2 s
Waku LightPush	78.6 KB (+0.02% 🔺)	1.6 s (+0.02% 🔺)	8.6 s (+61.45% 🔺)	10.2 s
History retrieval protocols	79.1 KB (-0.01% 🔽)	1.6 s (-0.01% 🔽)	5 s (+5.96% 🔺)	6.6 s
Deterministic Message Hashing	7.31 KB (0%)	147 ms (0%)	278 ms (-75.35% 🔽)	424 ms

[danisharora099]

While I would prefer we keep it as much as possible for now considering how unstable the nodes infrastructure is for light clients (high churn), RFC is indeed the source of truth.
Please make the shot here in whatever direction you think is the best!

[weboko]

@danisharora099 I think it should be fine. We have renewal in place so it should be essentially the same as current behavior and worst case is that 2 of the nodes are broken.

Next improvement would be to enable auto retrials on sent messages in LightPush.