feat: add MiniMax M3 as a cloud AI provider

[M3NT1]

What

This PR is a re-port of the MiniMax M3 provider integration onto the current main (v2.0.3) base, which refactored the provider system into LLMProviderID with separate cases for chatGPT, claude, openAICompatible, and local. It also brings over the model-management + usage-analytics layer that was in the original PR #320.

The user's request:

"Sync the upstream changes into the repo while keeping our MiniMax changes intact. And while you're at it, add a model picker that fetches live, a dashboard for provider/model usage, and a badge on each card."

Why a new branch?

PR #320 was opened against the v2.0.0 base. Since then upstream has landed 163 commits — including a major provider refactor that split chatGPTClaude into separate chatGPT / claude cases, introduced openAICompatible and local cases, added LLMProviderSetupPreferences, and rebuilt onboarding/settings around a new ProviderSetupState machine.

A straight rebase produced 21 conflicting files; rather than fight through them, this branch is a clean re-application onto the v2.0.3 base so both upstream's evolution and our MiniMax additions land together. #320 should be closed in favour of this PR.

Changes

Core/AI/LLMTypes.swift — added .minimax case to LLMProviderID with analytics + display label.

Core/AI/MiniMaxProvider*.swift (8 new files) — the provider itself, modelled on OllamaProvider but with Bearer auth against https://api.minimax.io/v1 and a 1M-token context window. Five files: main + Networking, Reasoning (strips <think>…</think> defensively), Summaries, Transcription. Plus MiniMaxModelPreference, MiniMaxPromptPreferences, and Utilities/MiniMaxAPIHelper.

Core/AI/ModelCatalog.swift (new) — actor that fetches the current model list from each provider's API on the fly:

• GET /v1/models for MiniMax (OpenAI-compatible)
• GET /v1beta/models?key=… for Gemini (filtered to vision-capable)
• GET {baseURL}/api/tags for Ollama, GET /v1/models for LM Studio
• Curated lists for ChatGPT/Claude CLIs (no public list endpoint)
• 24h cache in UserDefaults. On failure, falls back to a curated list with a banner in the UI.

Core/AI/LLMService.swift — new makeMiniMaxProvider() wired into both makeBatchProvider and makeTextProvider switch statements. A new providerModelId(for:) helper centralises the per-provider model lookup. A new attachMeta helper stamps every card before persistence.

Core/Analytics/ProviderStatsCalculator.swift (new) — pure data layer that turns the local TimelineCard list into a ProviderStatsSummary:

• byProvider: [String: Int] (always includes all known providers, even at zero)
• byModel: [(providerID, modelID, count)]
• daily: [DailyStat] (last 365 days)
• weekly: [WeeklyStat] (last 26 weeks)
• monthly: [MonthlyStat] (last 12 months)
• heatmapDays: [HeatmapDay] (last 365 days, intensity 0…1)

Views/UI/Dashboard/ProviderStatsView.swift (new) — self-contained dashboard surface with:

• 4 stat tiles (total cards, active providers, peak day, tracking since)
• Provider bar chart + percentages
• Per-(provider, model) breakdown
• GitHub-style 365-day heatmap (5 intensity buckets, 7×N grid)
• Daily / weekly / monthly bar chart with Week / Month / Quarter / Year range picker
• Manual refresh button

Storage layer — provider_id + model_id columns on timeline_cards with ALTER TABLE migration + composite index. TimelineCard and TimelineCardShell gained matching fields. New StorageManager.fetchAllTimelineCards() powers the dashboard.

Onboarding — 6th provider card on the comparison screen (sparkles icon, "MiniMax M3", positioned between OpenAI-compatible and Local). New 4-step setup flow: get-key → enter-key → test-connection → complete. Reuses the existing apiKey state field that the input view binds to.

Settings → Providers tab — MiniMax section with model picker (live or curated), keychain API-key field, test-connection button that re-opens the provider-setup modal. Prompt-override plumbing added for symmetry with the other providers.

Per-card badge — CanvasActivityCard (and equivalent list/grid renderers) now show a small text pill below the time label, e.g. MiniMax · MiniMax-M3, Claude · Sonnet 4.5, Local · qwen3-vl:4b. Falls back to "Unknown" when the card predates the schema migration.

ChatCLIProvider — gained a defaultModel: String? parameter that overrides the previously hard-coded "sonnet" / "haiku" / "gpt-5.4" strings. LLMService.makeChatCLIProvider reads llmCodexModel / llmClaudeModel from UserDefaults and passes them through.

Privacy

• The MiniMax API key is stored in the macOS Keychain, not UserDefaults
• Screenshots are sent to https://api.minimax.io/v1/chat/completions and nowhere else
• The provider-stats dashboard reads only the local SQLite DB — no network, no screenshot content, no PII
• Failure paths log HTTP status + body up to 64 KB via LLMLogger, but no PII

Build verification

xcodebuild -project Dayflow.xcodeproj -scheme Dayflow \
  -configuration Release -destination 'platform=macOS' \
  CODE_SIGN_IDENTITY="-" CODE_SIGNING_REQUIRED=NO CODE_SIGNING_ALLOWED=NO \
  -quiet build
# ** BUILD SUCCEEDED **

The pre-existing Swift 6 / Sendable warnings in ChatCLIProcessRunner.swift, MiniMaxProvider+Transcription.swift, etc. are unchanged from upstream and not introduced by this PR.