feat: allow specifying reasoning level for Slack integration

[RSO]

Summary

• Adds an optional Reasoning Level selector to the Slack integration settings UI, shown conditionally when the selected model supports reasoning_effort (detected via supported_parameters)
• Stores the reasoning effort alongside the model slug in the platform_integrations metadata JSONB column
• Passes the configured reasoning effort through to the OpenRouter API via the reasoning.effort field when the Slack bot processes messages

Changes

Service layer (slack-service.ts)

• updateModel() now accepts an optional reasoningEffort parameter and persists it in metadata
• Renamed getModel() → getModelConfig() returning a SlackModelConfig with both modelSlug and reasoningEffort

tRPC router (slack-router.ts)

• getInstallation response includes reasoningEffort (validated via ReasoningEffortSchema)
• updateModel mutation accepts optional reasoningEffort input
• Audit log includes reasoning effort when set

Bot engine (run-bot.ts)

• BotRunInput accepts optional reasoning: OpenRouterReasoningConfig
• Reasoning config is included in the request body when present

Slack bot runtime (slack-bot.ts)

• Reads reasoning effort from getModelConfig() and builds an OpenRouterReasoningConfig
• Passes it to runBot() via the new reasoning parameter

UI (SlackIntegrationDetails.tsx)

• Conditionally renders a Reasoning Level <Select> (None / Low / Medium / High) when the model supports it
• Automatically clears reasoning effort when switching to a model that doesn't support it
• Saves model + reasoning effort together in a single mutation