.Net: Update AI Clients to defend against URL injection attacks

[rogerbarreto]

Motivation and Context

This change adds security validation tests for URL injection attempts in various client constructors across the Google and Pinecone connectors. It helps prevent potential security vulnerabilities where malicious URLs could be injected through location or environment parameters.

Description

Added URL injection validation tests for:

• GeminiChatCompletionClient
• GeminiTokenCounterClient
• VertexAIEmbeddingClient
• PineconeClient

Each test suite includes:

1. Validation against common URL injection patterns (e.g., malicious protocols, script injection)
2. Verification of valid hostname/environment patterns
3. Consistent testing approach using xUnit Theory tests

The tests ensure that:

• Malicious URL attempts are caught and throw ArgumentException
• Valid location/environment names are accepted without throwing exceptions
• Constructor parameter validation works as expected

This change improves the security posture of the connectors by ensuring proper input validation and maintaining consistent validation behavior across different client implementations.

Tests

• Added new test files and test cases
• All tests pass locally
• No breaking changes introduced
• Maintains existing validation patterns