Fix issue #113: Prevent PaddingOracleAttacker from running endlessly

[ic0ns]

Summary

• Add early termination conditions to prevent endless execution when testing unresponsive servers
• Add maximum runtime limit to ensure the probe completes within reasonable time
• Improve error handling to continue testing other cipher suites when individual tests fail

Fixes #113

This PR addresses the issue where PaddingOracleAttacker can run for hours while producing thousands of warnings when testing certain targets.

Changes Made

PaddingOracleAttacker.java

• Added MAX_CONSECUTIVE_FAILURES constant (20) - stops testing after 20 consecutive fingerprint extraction failures
• Added MAX_FAILURE_RATE constant (0.75) - stops testing when more than 75% of attempts fail (after at least 10 attempts)
• Added failure tracking logic in createVectorResponseList() method
• Added informative logging about failure statistics

PaddingOracleProbe.java

• Added MAX_PROBE_RUNTIME_MS constant (20 minutes) - ensures probe completes in reasonable time
• Added runtime checks before processing each cipher suite and during extended evaluation
• Added try-catch blocks to handle individual test failures gracefully without stopping the entire probe

PaddingOracleProbeTest.java

• Added unit tests for the probe's basic functionality and configuration

Test Plan

• Code compiles successfully
• Code formatting applied with spotless
• Added unit tests for probe configuration
• Manual testing with problematic servers (requires access to specific test targets)

Notes

The fix uses a two-pronged approach:

1. Failure detection: Stops early when too many consecutive failures occur or the overall failure rate is too high
2. Time limit: Enforces a maximum runtime to prevent the probe from running indefinitely

This ensures the scanner remains responsive even when testing servers with connectivity issues or other problems that prevent successful fingerprint extraction.