[BUG] Limit for autoinstrumentation init container calculated wrong

[florianmutter]

Agent Environment

Agent 7.63.3 - Commit: eab078f2e7 - Serialization version: v5.0.141 - Go version: go1.23.5

Describe what happened:
Datadog webhook adds init containers with invalid resource spec. CPU limit can be smaller than request which is not allowed. This happens if there are multiple container in a pod where the sum all limits is smaller than the sum of all requests (e.g. because on container does not have a limit at all).

Describe what you expected:
Datadog Admission controller does not create invalid spec.

Steps to reproduce the issue:
Create a pod with two containers like this:

apiVersion: apps/v1
kind: Deployment
metadata:
  name: demo
spec:
  template:
    spec:
      containers:
        - name: container1
          resources:
            limits:
              cpu: 1000m
              memory: 512Mi
            requests:
              cpu: 1000m
              memory: 512Mi
        - name: container2
          resources:
            limits:
              memory: 128Mi
            requests:
              cpu: 50m
              memory: 128Mi

Additional environment details (Operating System, Cloud provider, etc):
GKE.

The code that does this calculation wrong is probably here: https://github.com/DataDog/datadog-agent/blob/main/pkg/clusteragent/admission/mutate/autoinstrumentation/auto_instrumentation.go#L288

Seems to have been introduced with #30266

[florianmutter]

Workaround is to add something like

env:
  - name: DD_ADMISSION_CONTROLLER_AUTO_INSTRUMENTATION_INIT_RESOURCES_CPU
    value: 100m
  - name: DD_ADMISSION_CONTROLLER_AUTO_INSTRUMENTATION_INIT_RESOURCES_MEMORY
    value: 100Mi

for the cluster agent.