Added memory requests to be more realistic.

This makes Kubernetes make better choices about where
to schedule the pods, and communicates to the administrators
about the minimum sensible resource requirements.

On a single user Mastodon instance on a three node Kubernetes
after a week of so use we get these memory uses per pod:
```
tero@arcones:~$ kubectl top pods -n mastodon
NAME                                           CPU(cores)   MEMORY(bytes)
mastodon-elasticsearch-coordinating-0          6m           403Mi
mastodon-elasticsearch-coordinating-1          28m          189Mi
mastodon-elasticsearch-data-0                  10m          1432Mi
mastodon-elasticsearch-data-1                  5m           1513Mi
mastodon-elasticsearch-ingest-0                6m           418Mi
mastodon-elasticsearch-ingest-1                6m           396Mi
mastodon-elasticsearch-master-0                24m          466Mi
mastodon-elasticsearch-master-1                10m          221Mi
mastodon-postgresql-0                          12m          276Mi
mastodon-redis-master-0                        16m          37Mi
mastodon-redis-replicas-0                      7m           34Mi
mastodon-sidekiq-all-queues-549b4bb7b4-zvj2m   266m         499Mi
mastodon-streaming-78465f778d-6xfg2            1m           96Mi
mastodon-web-774c5c94f9-f5bhz                  22m          418Mi
```

Hence we make the following adjustments to Bitnami defaults:
- `mastodon-elasticsearch-coordinating`: `256Mi->512Mi`
- `mastodon-elasticsearch-data`: The default `2048Mi` is ok.
- `mastodon-elasticsearch-master`: `256Mi->512Mi`
- `mastodon-redis-master`: `0->56Mi`
- `mastodon-redis-replicas`: `0->56Mi`
- `mastodon-postgresql`: `256->384Mi`

And for Mastodon defaults:
- `mastodon-sidekiq-all-queues`: `0->512Mi`
- `mastodon-streaming`: `0->128Mi`
- `mastodon-web`: `0->512Mi`

The original idea of keeping these requests zero is a good default when
minimal requirements are unknown. However, from a single user node
we get minimal requirements and having the limits as zero only leads
to trouble for people.
Of course the system requirements will change over time, but they
are chiefly expected to go upwards.

Author: keskival

values.yaml

@@ -82,15 +82,15 @@ mastodon:
     # -- (Sidekiq Container) Security Context for all Pods, overwrites .Values.securityContext
     securityContext: {}
     # -- Resources for all Sidekiq Deployments unless overwritten
-    resources: {}
-    # -- Affinity for all Sidekiq Deployments unless overwritten, overwrites .Values.affinity
-    affinity: {}
+    resources:
+      requests:
+        cpu: 250m
+        memory: 512Mi
     # limits:
     #   cpu: "1"
     #   memory: 768Mi
-    # requests:
-    #   cpu: 250m
-    #   memory: 512Mi
+    # -- Affinity for all Sidekiq Deployments unless overwritten, overwrites .Values.affinity
+    affinity: {}
     workers:
     - name: all-queues
       # -- Number of threads / parallel sidekiq jobs that are executed per Pod
@@ -162,13 +162,13 @@ mastodon:
     # -- (Streaming Container) Security Context for Streaming Pods, overwrites .Values.securityContext
     securityContext: {}
     # -- (Streaming Container) Resources for Streaming Pods, overwrites .Values.resources
-    resources: {}
+    resources:
+      requests:
+        cpu: 250m
+        memory: 128Mi
     # limits:
     #   cpu: "500m"
     #   memory: 512Mi
-    # requests:
-    #   cpu: 250m
-    #   memory: 128Mi
   web:
     port: 3000
     # -- Number of Web Pods running
@@ -180,13 +180,13 @@ mastodon:
     # -- (Web Container) Security Context for Web Pods, overwrites .Values.securityContext
     securityContext: {}
     # -- (Web Container) Resources for Web Pods, overwrites .Values.resources
-    resources: {}
+    resources:
+      requests:
+        cpu: 250m
+        memory: 512Mi
     # limits:
     #   cpu: "1"
     #   memory: 1280Mi
-    # requests:
-    #   cpu: 250m
-    #   memory: 768Mi
 
   metrics:
     statsd:
@@ -231,6 +231,18 @@ elasticsearch:
   # @ignored
   image:
     tag: 7
+  coordinating:
+    resources:
+      requests:
+        memory: 512Mi
+  ingest:
+    resources:
+      requests:
+        memory: 512Mi
+  master:
+    resources:
+      requests:
+        memory: 512Mi
 
 # https://github.com/bitnami/charts/tree/master/bitnami/postgresql#parameters
 postgresql:
@@ -253,6 +265,10 @@ postgresql:
     # you can also specify the name of an existing Secret
     # with a key of password set to the password you want
     existingSecret: ""
+  primary:
+    resources:
+      requests:
+        memory: 384Mi
 
 # https://github.com/bitnami/charts/tree/master/bitnami/redis#parameters
 redis:
@@ -268,6 +284,14 @@ redis:
   # with a key of redis-password set to the password you want
   # auth:
     # existingSecret: ""
+  master:
+    resources:
+      requests:
+        memory: 56Mi
+  replica:
+    resources:
+      requests:
+        memory: 56Mi
 
 # @ignored
 service: