If no resource_requirements are set by the user, the deployment fails

[rooftopcellist]

Version
main branch

Describe the bug

If no resource_requirements are set by the user, the deployment fails. The user should be able to either configure these, or leave them blank.

If they are not specified, redis never comes up, and the rest of the deployment is botched as well.

Error seen:

$ oc get pods
oc logs NAME                                                  READY   STATUS     RESTARTS       AGE
galaxy-api-74786cb9c4-5xbkg                           0/1     Pending    0              8m4s
galaxy-content-747df6668c-2cdrt                       0/1     Init:0/1   2 (114s ago)   8m42s
galaxy-content-747df6668c-9sjz5                       0/1     Init:0/1   2 (64s ago)    8m42s
galaxy-operator-controller-manager-657f4b5466-xwpq9   2/2     Running    0              10m
galaxy-postgres-13-0                                  1/1     Running    0              9m11s
galaxy-redis-54b6b744db-rgrx5                         0/1     Pending    0              8m25s
galaxy-web-5799ff48f9-pntnf                           0/1     Init:0/1   0              8m58s
galaxy-worker-5896c85676-lw6zc                        0/1     Init:0/2   2 (116s ago)   8m33s
galaxy-worker-5896c85676-vwjsm                        0/1     Init:0/2   2 (65s ago)    8m33s

$ oc get pvc
NAME                            STATUS    VOLUME                                     CAPACITY   ACCESS MODES   STORAGECLASS    AGE
galaxy-file-storage             Bound     pvc-726d28e2-489d-4b1c-87c3-e5d8f092db85   10Gi       RWX            nfs-local-rwx   8m53s
galaxy-redis-data               Pending                                                                        standard-ssd    8m33s
postgres-galaxy-postgres-13-0   Bound     pvc-ebb24b2b-ae4f-4c45-8288-e5238584c6a6   8Gi        RWO            standard-ssd    9m17s

Additional context

This is a side-effect of having settings like resource_requirements nested under each component of the app. (api.resource_requirements for example).

This is how we solved that problem in the eda-server-operator:
https://github.com/ansible/eda-server-operator/blob/main/roles/eda/tasks/combine_defaults.yml
I think we should do something similar for the galaxy-operator.