Update example performance doc (#1636)

Original values don't make sense from a performance point of view as
they are lower than the default values. Also high number of threads
combined with bottleneck on QPS has the opposite effect as objects are
re-queued and wait in the workqueue with exponentially increasing
timeout.
Fixed ref to old image registry.

Author: enarha

docs/performance.md

@@ -19,9 +19,9 @@ spec:
     template:
         spec:
             containers:
-                - image: gcr.io/tekton-releases/github.com/tektoncd/chains/cmd/controller:v0.20.0
+                - image: ghcr.io/tektoncd/chains/controller-92006fd957c0afd31de6a40b3e33b39f:v0.26.2
                   args:
                     - --threads-per-controller=32
-                    - --kube-api-burst=2
-                    - --kube-api-qps=3
+                    - --kube-api-burst=100
+                    - --kube-api-qps=50
 ```