Replace linear IP blocklist scan with indexed trie lookup

[Mishenevd]

IPList.containsAddress() scanned every stored subnet/host linearly (O(n)).

At 300k+ entries this costs ~13.7ms per lookup;

isIpBlocked runs it up to 4x per request.

Switched to inet.ipaddr's built-in DualIPv4v6Tries, giving O(bits) containment lookups (~1.2us at the same scale):

A compact binary trie (aka compact binary prefix tree, or binary radix trie)... entry length is capped at 128 bits for IPv6 and 32 bits for IPv4. That makes lookup a constant time operation... construction is O(n), in linear proportion to the number of added elements.

Public API and matches() semantics unchanged; all existing IPListTest cases pass unmodified.

Summary by Aikido

Security Issues: 0	Quality Issues: 0	Resolved Issues: 0

⚡ Enhancements

• Replaced linear IP set with DualIPv4v6Tries for faster lookups

🔧 Refactors

• Refactored containsAddress to use trie elementContains call and preserved semantics

^{More info}

[codecov]

Codecov Report

❌ Patch coverage is 0% with 2 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
.../java/dev/aikido/agent_api/helpers/net/IPList.java	0.00%	2 Missing ⚠️

📢 Thoughts on this report? Let us know!