chore: bump version to v0.2.0 and promote CHANGELOG

[artcava]

Summary

Added

• Prompt request value objects for orchestrator-owned intent (#223): introduced PromptRequest and ImagePromptRequest records to carry prompt data from orchestrators to providers, making prompt intent a first-class concept decoupled from provider configuration.
• Role-keyed prompt options for FeedOrchestrator (#223): added PromptRole, PromptStepOptions, and FeedPromptOptions to model an ordered collection of prompt steps (Summary, ImagePromptDerivation, ImageGeneration) resolved by the orchestrator for each phase of the feed pipeline.
• Slot-scoped FeedOrchestratorContext (#223): new value object carrying FeedUrls and FeedPromptOptions per scheduled slot, enabling multiple FeedOrchestrator slots to run with independent feed sources and prompt configurations.
• ScheduledOrchestrationProfile.OrchestratorContextKey (#223): added a logical context key used at runtime to resolve the appropriate FeedOrchestratorContext for each slot.

Changed

• SenderPlugin DI optimized with Keyed registration
• AiProviderOptions optimized in a single Extension method
• BuildChatPayload centralized in AiServiceHelper (#244): eliminated identical private BuildChatPayload implementations across AzureFoundryService, OpenAiService, DeepSeekService, and PerplexityService; a single internal static object BuildChatPayload(string text, PromptRequest request, string modelName) method now lives in AiServiceHelper, with each provider delegating to it and passing its provider-specific model name.
• Prompt ownership aligned with architectural boundaries (#223): prompt composition is now owned by orchestrators, while providers focus solely on execution and connectivity/capability settings.
• Execution-oriented provider contracts (#223): text and image providers now accept dedicated request objects (PromptRequest / ImagePromptRequest) instead of reading prompt fields from provider options.
• FeedOrchestrator refactored for fan-out with slot context (#223): the orchestrator uses role-keyed prompt steps, generates the base summary once for the widest sender, reuses the Summary step for re-summarisation of narrower senders, and shares a single image generation across all configured senders in the slot.
• Configuration migrated to FeedSlotContexts__{key} sections (#223): feed URLs and prompt options are now defined per logical slot context, replacing the previous global configuration and enabling per-slot tuning.
• OrchestratorFactory updated to resolve keyed context (#223): the factory now selects the correct FeedOrchestratorContext based on OrchestratorContextKey while preserving centralised orchestrator selection and existing reflection-based instantiation.
• upgraded .net version to 10 LTS from 8 LTS ([#169]): also upgraded some library and revised all documentation
• Substituted MaskUrlTelemetryInitializer with MaskUrlTelemetryProcessor to mask access_token in Facebook Graph API HTTP dependencies
• upgraded .net version to 10 LTS from 8 LTS ([#169]): also upgraded some library and revised all documentation.

Removed

• Prompt intent from provider options (#223): prompt-related fields such as templates, temperature, output/token limits, and input labels have been removed from provider option models to avoid mixing orchestration behaviour into provider configuration.
• Global configuration assumption for FeedOrchestrator (#223): the previous single shared feed URL list and prompt configuration for all FeedOrchestrator slots has been removed in favour of per-slot contexts.

[github-actions]

⚠️ This PR does not reference an approved issue. Please link to an issue using "Closes #123" in the description.