[telemetry] Add retry logic for partial telemetry push failures

[akgitrepos]

Description

Implements retry logic for partial telemetry push failures in TelemetryPushClient. When the telemetry service returns a partial success response (numProtoSuccess < totalEvents), the client now automatically retries up to 3 times with exponential backoff (1s → 2s → 4s, max 10s).

This addresses the TODO comment in TelemetryPushClient.java.

Changes Made

• Added configurable maxRetries parameter (default: 3)
• Implemented exponential backoff strategy (aligned with DatabricksHttpRetryHandler)
• Added unit tests for retry behavior

Testing

• Added 2 new unit tests:
  • pushEvent_noRetryOnFullSuccess - Verifies no retry when all events succeed
  • pushEvent_retriesOnPartialSuccess - Verifies retry is triggered on partial failure
• All 14 TelemetryPushClient tests pass

Additional Notes

• Backward compatible - existing constructor still works with default retry count
• Idempotent - retries all events (safe for duplicates)

[akgitrepos]

@gopalldb @vikrantpuppala Hi, I've implemented retry logic for partial telemetry push failures which was one of the TODO item. Would appreciate a review. Thanks!