fix: filter stats mismatch and add bad peer check for light mode

[chaitanyaprem]

Description

While debugging filter issue, noticed that filter stats are mismatched because once a filter is removed it is still left in filtersubscriptions list in filterManager causing logs to be inconsistent.

Also added a minor bad peer removal functionality for light nodes based on dial error count. This may help improve lightnode connectivity and reduce failures.
Logic is simple:

1. Consider a peer as bad-peer if its dial error count is more than 3.
2. Keep checking for such bad peers and remove them every 1 minute.

[status-im-auto]

Jenkins Builds

Click to see older builds (3)

❔	Commit	#️⃣	Finished (UTC)	Duration	Platform	Result
✔️	ac9241b	#1	2024-10-15 13:58:10	~1 min	unknown	📄log
✔️	9c4ab85	#2	2024-10-15 14:07:56	~2 min	unknown	📄log
✔️	c4fd528	#5	2024-10-16 10:17:50	~2 min	unknown	📄log

❔	Commit	#️⃣	Finished (UTC)	Duration	Platform	Result
✔️	3546adf	#6	2024-10-18 10:49:18	~1 min	unknown	📄log
✔️	653e7b5	#7	2024-10-28 06:06:10	~2 min	unknown	📄log

[chaitanyaprem]

@siddarthkay not sure why the nix-build is failing, i have tried rerunning it but still fails. can you take a look?

[siddarthkay]

@chaitanyaprem : works now

[chaitanyaprem]

@chaitanyaprem : works now

Thanks a lot! was it some sort of temporary failure or something you had to fix?

[siddarthkay]

temp failure on linux-03 host

[richard-ramos]

LGTM. The only question i have is if these logs could be too noisy to have them at Info level

[chaitanyaprem]

LGTM. The only question i have is if these logs could be too noisy to have them at Info level

hmm, good point...will make them debug.