Move most of the remaining tests to Test Containers as well

[siri-varma]

Description

Finishes migrating the sdk-tests integration tests off the legacy dapr run CLI harness and onto Testcontainers (BaseContainerIT / DaprContainer), and strips the now-unused legacy setup from the build CI job. This removes the slow, flaky host-daprd path that was driving the build sb:3.5.x job toward its timeout.

What changed

Test migration (last legacy ITs → Testcontainers)

• Migrated the remaining CLI-based ITs: ActorReminderRecoveryIT, ActorReminderFailoverIT, WaitForSidecarIT, BindingIT (now Testcontainers Kafka + in-code bindings), and ActorSdkResiliencyIT (moved off ToxiProxyRun; still @Disabled).
• Added reusable mechanics to BaseContainerIT: startAppAndAttach (app subprocess + pinned-port daprd), restartApp / restartSidecar, waitForActorsReady, and newToxiproxy (Testcontainers ToxiProxy for network-fault injection — replaces the host-installed binary; same toxics, now hermetic and version-pinned).
• Centralized shared backing services in SharedTestInfra: one Docker Network + JVM-singleton Redis / Zipkin / Mongo / Kafka.

Actor IT flakiness fix — shared control plane (DNS-alias collision)

• Root cause: every actor IT attaches daprd to the JVM-wide shared Network. The builder used withReusablePlacement(true); with Testcontainers reuse disabled (as on CI), DaprContainer.configure() auto-creates a placement and scheduler per daprd instance — each claiming the placement / scheduler network alias and never stopped (no stop() override). Multiple containers answering one alias make Docker DNS round-robin daprd to an arbitrary/empty control plane, so multi-sidecar failover (ActorReminderFailoverIT) and post-restart reminder recovery (ActorReminderRecoveryIT) hit a placement that never saw the actor-host registrations → did not find address for actor / reminders not resumed.
• Fix: JVM-singleton placement() / scheduler() in SharedTestInfra on the shared network, wired into every daprd via withPlacementContainer / withSchedulerContainer. Exactly one container answers each alias, so DNS resolves deterministically; this also removes the per-daprd control-plane container leak.

CI cleanup

• Dropped Dapr CLI + dapr init, host Kafka, and host ToxiProxy setup from the build job.
• Removed the dead dapr run harness, binding YAMLs, and local-test.yml.

Issue reference

Closes #1522

Test results

• Actor package: green — ActorReminderRecoveryIT (4/4) and ActorReminderFailoverIT (1/1), which failed deterministically before the fix, now pass both in isolation and in the full suite.
• tracing.http.TracingIT: fixed. Its Spring @Bean app eagerly called the legacy OpenTelemetry.createOpenTelemetry(SERVICE_NAME) overload, which probes a fixed 127.0.0.1:9411 for Zipkin readiness. After this migration Zipkin runs as a Testcontainer on a random mapped host port, so the probe threw ConnectException, the app subprocess crashed before binding its port, and startAppAndAttach reported "connection refused". Switched the bean to the explicit-endpoint overload that skips the probe (the app exports no spans the test asserts on — the validated calllocal/tracing-http-it/sleep span is emitted by the sidecar over the container network). Passes in isolation.
• Full sdk-tests suite (local): 7 → 1 error. All four actor errors resolved by the control-plane fix; TracingIT.http resolved by the bean fix above.
• 1 remaining local failure (spring.messaging.DaprSpringMessagingIT) is a topic-subscription readiness timeout. It reproduces even in isolation and is independent of this PR — it is a standalone @SpringBootTest that manages its own container and touches none of the code changed here. It appears environment-sensitive on a local macOS Docker host (uses host.testcontainers.internal); CI on this PR is the arbiter for whether it affects the target environment. Tracked as follow-up, not introduced here.

Checklist

• Code compiles correctly
• Created/updated tests
• Extended the documentation

[siri-varma]

@javier-aliaga / @cicoyle Please double check this change.

We are able to save 30 mins because of this change. It was taking 32 mins before now just takes 4 mins

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 77.06%. Comparing base (26b2327) to head (c7fdb76).

Additional details and impacted files

@@             Coverage Diff              @@
##             master    #1785      +/-   ##
============================================
+ Coverage     77.02%   77.06%   +0.04%     
  Complexity     2324     2324              
============================================
  Files           245      245              
  Lines          7198     7198              
  Branches        760      760              
============================================
+ Hits           5544     5547       +3     
+ Misses         1288     1285       -3     
  Partials        366      366              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[siri-varma]

@javier-aliaga need one review on this pr please