feat(host): add stream feature abstraction layer

[j-d-ha]

Problem Statement

Currently, input and output streams are accessed directly via RawInvocationData.Event and RawInvocationData.Response. This requires handlers to work with low-level stream abstractions rather than leveraging the framework's feature-based access pattern that already exists for other Lambda context data.

Proposed Solution

Create feature-based abstractions for input and output streams (IInputStreamFeature, IOutputStreamFeature) that follow the established pattern of type-safe feature access in the framework (similar to IEventFeature). This allows handlers to access streams through the abstraction layer while maintaining backward compatibility.

Acceptance Criteria

• Feature interfaces for input/output streams defined in AwsLambda.Host.Abstractions
• Default implementations provided in AwsLambda.Host
• Handler parameter binding support (via attribute or dependency injection)
• Raw streams still accessible via RawInvocationData for backward compatibility
• Update existing examples to demonstrate feature-based stream access
• Documentation/examples updated with new pattern

Implementation Notes

Key areas to modify:

• src/AwsLambda.Host.Abstractions/Core/Features/ - Add stream feature interfaces
• src/AwsLambda.Host/Features/ - Add default feature implementations
• src/AwsLambda.Host/Services/ - Integrate with feature binding/resolution
• Example projects - Update to demonstrate feature-based access