Closes #3673

Motivation and Context

Add Azure AI Foundry Memory Context Provider

Introduces Microsoft.Agents.AI.FoundryMemory, a new context provider package that enables agents to store
and retrieve long-term memories using Azure AI Foundry's managed memory service. The provider
automatically extracts memories from conversations and retrieves relevant memories to augment agent
context on each invocation, following the same pattern established by the Mem0 provider.

Features

• FoundryMemoryProvider implementing AIContextProvider with automatic memory extraction and retrieval
• Overrides ProvideAIContextAsync and StoreAIContextAsync following the updated AIContextProvider base class pattern
• Uses ProviderSessionState<State> for automatic session state management (same pattern as Mem0Provider)
• Scoped memory storage by configurable identifier (e.g., user ID) for multi-tenant scenarios via FoundryMemoryProviderScope
• EnsureMemoryStoreCreatedAsync() to create memory stores with specified chat and embedding models
• EnsureStoredMemoriesDeletedAsync() to clear memories for a scope
• WhenUpdatesCompletedAsync() to poll pending memory updates in parallel until completion
• Configurable options (FoundryMemoryProviderOptions) for max memories, update delay, context prompt, state key, and message filters
• Sample application (AgentWithMemory_Step04_MemoryUsingFoundry) demonstrating end-to-end usage with session serialization
• Unit tests for constructor validation
• Integration tests using agent-based API (skipped by default, require live Foundry service)

Contribution Checklist

• The code builds clean without any errors or warnings
• The PR follows the Contribution Guidelines
• All unit tests pass, and I have added new tests where possible
• Is this a breaking change? If yes, add [BREAKING] prefix to the title of the PR.