Fix #11: Show running Spark applications before event logs are uploaded

[idirze]

As described in the issue #11 , when Spark is configured with S3-backed event logs, there is a delay, controlled by spark.eventLog.rolling.maxFileSize, before event logs are uploaded. During this time:

• Spark History Server shows “No incomplete applications found!”
• Users cannot access the Spark UI for running applications

This PR improves Spark History incomplete appliactions page by:

• Detecting running Spark applications from Kubernetes pods
• Fetching live application info directly from running Spark UI endpoints
• Merging running apps with Spark History in progress results
• Supporting Spark 3.x and Spark 4.x
• Making the reverse proxy more resilient (gzip handling, redirects, client disconnects)

As a result, users can access Spark UI immediately, even before event logs are uploaded to S3.

📌 Related Issue

Fixes #11