fix: improve single-region monitoring scrapes and gzip metrics

[polinikita]

Summary

• use fixed shared Prometheus scrape jobs instead of one job per node
• make external-monitoring scrape address selection explicit: external monitors default to public validator IPs, with an override for private-network deployments
• gzip Prometheus metrics responses to reduce scrape bandwidth

Verification

• CARGO_TARGET_DIR=/tmp/starfish-verify-orch cargo test -p orchestrator -- --nocapture
• CARGO_TARGET_DIR=/tmp/starfish-verify-core cargo test -p starfish-core --no-run

[github-actions]

Package	Line Rate	Health
crates/starfish-core/src	40%	❌
crates/starfish-core/src/runtime	23%	❌
crates/orchestrator/src/client	7%	❌
crates/orchestrator/src/protocol	5%	❌
crates/starfish/src	3%	❌
crates/starfish-core/src/consensus	25%	❌
crates/starfish-core/src/core_thread	0%	❌
crates/orchestrator/src	18%	❌
Summary	34% (4260 / 12607)	❌