[Internal] Tests: Adds timeouts and polling loops for low-impact test reliability

[NaluTripician]

Summary

Adds [Timeout] attributes to prevent pipeline budget overruns and replaces fixed Task.Delay waits with bounded polling loops across multiple test files.

Changes

Timeout Attributes

• ClientTelemetryTests — [Timeout(300000)] (5 min) on all 12 test methods. Prevents hanging tests from consuming the entire 60-minute job budget.
• EndToEndTraceWriterBaselineTests — [Timeout(300000)] (5 min) on QueryAsync and TypedPointOperationsAsync.
• CosmosHttpClientCoreTests — [Timeout(120000)] (2 min) on retry tests + increased delay margins past policy thresholds.
• GlobalEndpointManagerTest — [Timeout(30000)] on EndpointFailureMockTest.

Polling Loops (replacing fixed delays)

• GlobalEndpointManagerTest — Polls for endpoint switch-back instead of fixed 3s sleep.
• BatchAsyncStreamerTests — Polls for semaphore count instead of fixed 2s sleep.
• PartitionControllerTests — Polls for lease release via Mock.Verify instead of fixed 100ms sleep.
• CosmosAuthorizationTests — Bounds the background refresh polling loop with a 20s timeout.

Tests Improved

Test	Failures (30 days)	Change
PointSuccessOperationsTest	9	Timeout added
EndpointFailureMockTest	2	Polling + Timeout
RetryTransientIssuesTestAsync	1	Timeout + margins
RetryTransientIssuesForQueryPlanTestAsync	1	Timeout + margins
ValidatesCongestionControlAsync	1	Polling loop
Controller_ShouldReleaseLease_IfObserverExits	1	Polling loop
+ 8 more tests	preventive	Timeout attributes

Impact

• ~14 flaky failures eliminated + pipeline budget overrun prevention
• Split from [Internal] Tests: Fixes test reliability across flaky and nightly tests #5643 for independent validation