fix(common): improve probes

There is a design mistake in the probe setup:

Liveness probe:
Stops the container on failure, this should happen AFTER readiness probe detects failure.

Readiness Probe:
Stops the service forwarding traffic. This should happen BEFORE the liveness probe detects failure

Startup Probe:
By having a tcp startup-probe, charts might pass CI when not fully working. or, worse, get passed-over to the readiness/liveness probes when the container is still doing maintenance work (such as plex database updates). Leading to it right-away being thrown-out by liveness/readyness

library/common/values.yaml

@@ -34,21 +34,21 @@ global:
     # -- Default probe timeouts
     probeTimeouts:
       liveness:
-        initialDelaySeconds: 10
-        periodSeconds: 10
-        timeoutSeconds: 5
-        failureThreshold: 5
-        successThreshold: 1
-      readiness:
-        initialDelaySeconds: 10
-        periodSeconds: 10
+        initialDelaySeconds: 20
+        periodSeconds: 15
         timeoutSeconds: 5
         failureThreshold: 5
         successThreshold: 2
+      readiness:
+        initialDelaySeconds: 20
+        periodSeconds: 10
+        timeoutSeconds: 5
+        failureThreshold: 3
+        successThreshold: 1
       startup:
         initialDelaySeconds: 10
         periodSeconds: 5
-        timeoutSeconds: 2
+        timeoutSeconds: 3
         failureThreshold: 60
         successThreshold: 1
     # -- Define a postgresql version for CNPG
@@ -200,7 +200,7 @@ workload:
               port: "{{ $.Values.service.main.ports.main.targetPort | default .Values.service.main.ports.main.port }}"
             startup:
               enabled: true
-              type: "tcp"
+              type: "{{ .Values.service.main.ports.main.protocol }}"
               port: "{{ $.Values.service.main.ports.main.targetPort | default .Values.service.main.ports.main.port }}"
 
 # -- Timezone used everywhere applicable