healthz endpoint returns "NOT OK" when `--events` or `--policy` is used (unless default events are enabled)

[hsandhu2309]

Description

The /healthz endpoint reports "NOT OK" when Tracee is started with custom events (e.g., --events anti_debugging) or any --policy, even though the server is running correctly. It only returns "OK" when the default event set is used — which includes the internal heartbeat_event.

This makes the health check unreliable for non-default configurations.

---

Steps to Reproduce

1. Build Tracee in a clean environment:

   make -f builder/Makefile.tracee-make ubuntu-shell
   make all

2. Start Tracee with healthz and a custom event:

   sudo ./dist/tracee --server healthz --events anti_debugging

3. Query the health endpoint:

   curl localhost:3366/healthz

   Result: NOT OK

4. Now start with default events:

   sudo ./dist/tracee --server healthz --events default

   curl localhost:3366/healthz

   Result: OK

5. Same failure occurs with any --policy flag:

   sudo ./dist/tracee --server healthz --policy mypolicy.json

   → NOT OK

---

Expected Behavior

/healthz should return "OK" as long as the Tracee server is running and responsive — regardless of event selection or policy usage.

---

Root Cause Analysis

The health status depends on the internal heartbeat_event , which calls SendPulse() → setHealth(true) and sets isHealthy = true.

• default events include heartbeat_event → health check passes
• Custom events (like anti_debugging) or policies do not include heartbeat_event → isHealthy remains false → "NOT OK"

Thus, healthiness is incorrectly tied to the presence of an internal diagnostic event rather than actual server liveness.

---

Environment

• Tracee version: main-0b8ac82ca
• OS: Ubuntu (via tracee-shell)
• Build: make all in ubuntu-shell

---

Suggested Fix

Decouple /healthz from heartbeat_event. The endpoint should reflect server readiness (e.g., listening, no critical init errors), not event subscription.

Possible approaches:

• Include heartbeat_event automatically when healthz is enabled
• OR: Manually add health function and then call heartbeat.SendPulse() (Not sure where to add this)

---

Additional Context

This breaks Kubernetes liveness probes and monitoring when using custom policies or focused event sets. Users expect healthz to indicate service availability.

---

Priority: High (affects observability in production)