fix: Close both apps to allow for process restart

[kamilogorek]

Currently when the main app is closed, or any node.js termination signal is sent (which is hijacked by closeWithGrace https://github.com/mcollina/close-with-grace?tab=readme-ov-file#api), an adminApp instance will keep it hostage and won't allow ec2 to restart everything correctly.

This can be tested by updating /health endpoint to call app.close() and observing that process is still running due to adminApp server being bound.

This will hopefully kill server instantly and won't let other requests come in to this instance as observed in our ec2 logs now.

Sidenote: why do we expose prometheus metrics on a separate port/app and not just /metrics endpoint within same app instead? @soedirgo

[coveralls]

Pull Request Test Coverage Report for Build 13160373616

Details

• 0 of 7 (0.0%) changed or added relevant lines in 2 files are covered.
• 1 unchanged line in 1 file lost coverage.
• Overall coverage decreased (-0.06%) to 75.116%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
src/server/admin-app.ts	0	1	0.0%
src/server/server.ts	0	6	0.0%

Files with Coverage Reduction	New Missed Lines	%
src/server/server.ts	1	0.0%

Totals
Change from base Build 13149849112:	-0.06%
Covered Lines:	4848
Relevant Lines:	6368

---

💛 - Coveralls

[pcnc]

Sidenote: why do we expose prometheus metrics on a separate port/app and not just /metrics endpoint within same app instead

I think it's so that we don't expose any type of metrics (not aware of the types of metrics we're exposing) - but eitherway those aren't being scraped either atm