Flaky Unit tests: fail when cluster is running, pass when cluster is deleted

[Horiodino]

Description

These tests fail when a real cluster is running locally but pass when the cluster is deleted.
This makes them flaky in local development.

Failure

=== SKIP: internal/controllers TestRegisterWithManager_OpAMPBridge (0.00s)
    opampbridge_controller_test.go:159: this test requires a real cluster, otherwise the GetConfigOrDie will die

=== SKIP: internal/controllers TestRegisterWithManager (0.00s)
    reconcile_test.go:1188: this test requires a real cluster, otherwise the GetConfigOrDie will die

=== FAIL: cmd/otel-allocator/internal/collector Test_runWatch/pod_add (30.75s)
W0711 12:25:54.585053  487575 shared_informer.go:591] resyncPeriod 2ms is too small. Changing it to the minimum allowed value of 1s
    collector_test.go:201: 
        	Error Trace:	/home/kalki/Documents/opentelemetry-operator/cmd/otel-allocator/internal/collector/collector_test.go:204
        	Error:      	"map[test-pod1:test-pod1]" should have 3 item(s), but has 1
    collector_test.go:201: 
        	Error Trace:	/home/kalki/Documents/opentelemetry-operator/cmd/otel-allocator/internal/collector/collector_test.go:205
        	Error:      	Not equal: 
        	           	expected: map[string]*allocation.Collector{"test-pod1":..., "test-pod2":..., "test-pod3":...}
        	           	actual  : map[string]*allocation.Collector{"test-pod1":...}
        	Diff:
        	--- Expected
        	+++ Actual
        	@@ -1,14 +1,4 @@
        	-(map[string]*allocation.Collector) (len=3) {
        	+(map[string]*allocation.Collector) (len=1) {
        	 (string) (len=9) "test-pod1": (*allocation.Collector)({
        	  Name: (string) (len=9) "test-pod1",
        	-  NodeName: (string) (len=9) "test-node",
        	-  NumTargets: (int) 0
        	- }),
        	- (string) (len=9) "test-pod2": (*allocation.Collector)({
        	-  Name: (string) (len=9) "test-pod2",
        	-  NodeName: (string) (len=9) "test-node",
        	-  NumTargets: (int) 0
        	- }),
        	- (string) (len=9) "test-pod3": (*allocation.Collector)({
        	-  Name: (string) (len=9) "test-pod3",
        	  NodeName: (string) (len=9) "test-node",

    collector_test.go:201: 
        	Error Trace:	/home/kalki/Documents/opentelemetry-operator/cmd/otel-allocator/internal/collector/collector_test.go:201
        	Error:      	Condition never satisfied
        	Test:       	Test_runWatch/pod_add

=== FAIL: cmd/otel-allocator/internal/collector Test_runWatch (30.86s)

DONE 1797 tests, 2 skipped, 2 failures in 34.799s
make: *** [Makefile:298: test] Error 1

Reproduce

make test

Branch and commit

Branch: main
Latest commit:

commit ef60b5746a3868e1e26691fac8dfb9681e1fe5d8 (HEAD -> main, origin/main, origin/HEAD)
Author: Mikhail Chinkov <mikhail.chinkov@coralogix.com>
Date:   Thu Jul 10 18:54:00 2025 +0200

    fix/target-allocator: check CRDs availability in the cluster (#4118)

[iblancasa]

@cazorla19 can you take a look into this?

[swiatekm]

This is addressed by #4180. It doesn't have anything to do with envtest clusters or @cazorla19's PR, but rather with fake Kubernetes clients leading to race conditions with informers in certain circumstances.