Fix/observability gaps

[renatoaf]

• Add tracing span for connection writes (we're noticing some high times in some of our games, we might want to add a timeout here in the future)
• Properly wrap errors when low-level connection writes fail (they are currently failing with a PIT-000 code in the metrics)
• Add error logging on some spots we were previously missing

[felipejfc]

@renatoaf idk if it was with these changes, but the unit tests are failing, I've already re-ran they 3 times but no luck. can you check before merging?

[coveralls]

Pull Request Test Coverage Report for Build 9668461825

Details

• 44 of 53 (83.02%) changed or added relevant lines in 4 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.2%) to 61.991%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
agent/agent.go	41	43	95.35%
cluster/nats_rpc_server.go	2	4	50.0%
service/handler.go	0	5	0.0%
<!--	Total:	44	53	83.02%	-->

Totals
Change from base Build 8848487306:	0.2%
Covered Lines:	4875
Relevant Lines:	7864

---

💛 - Coveralls

[renatoaf]

@renatoaf idk if it was with these changes, but the unit tests are failing, I've already re-ran they 3 times but no luck. can you check before merging?

Yep. Fixed the test and added a new one covering this scenario (closed agent after a write fails)

[coveralls]

Pull Request Test Coverage Report for Build 9668735601

Details

• 44 of 53 (83.02%) changed or added relevant lines in 4 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.2%) to 61.991%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
agent/agent.go	41	43	95.35%
cluster/nats_rpc_server.go	2	4	50.0%
service/handler.go	0	5	0.0%
<!--	Total:	44	53	83.02%	-->

Totals
Change from base Build 8848487306:	0.2%
Covered Lines:	4875
Relevant Lines:	7864

---

💛 - Coveralls