Captain bridge: clock, map marker, action stacks

name: Bridge clock and stacks overview: "Add wall-clock sim pacing with a three-way decision-attention mode (default: never pause), a distinct you-are-here star-map marker including in-transit progress, and a per-actor captain intent stack/state machine so multiple dock decisions can be queued and executed in order." todos:

• id: clock-model

content: Add DecisionAttention + SimSpeedScale; wall-clock pacer in LiveSession; default RunAlways; snapshot/HTTP/UI controls status: completed

• id: map-marker

content: Ship pose on CaptainBridgeModel + StarMapControl marker (dock + in-transit interpolation) status: completed

• id: intent-stack

content: CaptainIntentStack SM for Calypso; compound depart/reposition; bridge+session stack APIs; migrate bunker/lift into named steps status: completed

• id: cleanup-debug

content: Remove debug-cc3657 agent instrumentation from tramp/StubPhases status: completed isProject: false

Current baseline

• Time runs at max CPU in 24h outer pulses of 1h ticks (`CampaignRunner.LiveSession.RunAsync` + `CampaignPulse`). No wall-clock pacing.
• Pause policy is `CaptainPauseMode`: player default is `UntilDecision` (hard gate). UI: Step / Continue / To horizon / Pause next day only.
• Orders already FIFO in `PlayerOrderQueue`; agent drains one per tick. No named multi-step SM; bunker/lift requeues the same verb.
• Map (`StarMapControl`): hubs + route highlight; no ship marker. Position is voyage text + hub-level CurrentHubId.

1. Simulation speed + decision attention

Model (bridge-owned, exposed on snapshot + session commands):

flowchart LR
  subgraph clock [BridgeClock]
    Speed[SimSpeedScale]
    Attention[DecisionAttention]
  end
  Speed --> Pacer[WallClockPacer]
  Attention -->|RunAlways| Free[No gate]
  Attention -->|SoftSlow| Throttle[Cap rate while NeedsAttention]
  Attention -->|HardPause| Gate[Existing dayGate wait]
  Pacer --> Pulse[CampaignPulse hour ticks]

• `DecisionAttention` (replace sticky misuse of Continue/Resume as the only policy):
• RunAlways — default: never hard-pause for dock decisions; time flows at selected speed; stack stays fillable.
• SoftSlow — while NeedsPlayerDecision() (or non-empty blocking stack), multiply speed by a fixed factor (e.g. 0.1×) instead of blocking.
• HardPause — today’s UntilDecision gate (Continue/Step wake).
• Keep EveryDay / Step 1d / Pause-next-day as one-shot overlays that restore to the sticky DecisionAttention after the wait (same pattern as today restoring UntilDecision).
• `SimSpeedScale`: continuous control mapped to wall delay between sim hours (or between outer day pulses at the fast end):
• Slow end: ~1 real minute ≈ 1 game hour → ~60s delay per hour tick (or batch with equivalent average).
• Fast end: no delay — current CPU max; outer pulse remains ~1 game day per pulse (already hoursPerPulse = 24).
• UI: slider + presets (Crawl / Play / Fast / Max). Persist in bridge prefs for the process; expose via HTTP (setClock / snapshot fields) so agents can set speed without UI.
• Implement pacing in LiveSession.RunAsync between hour/day steps with cancellable delay; do not sleep inside sim.AdvanceAsync. When HardPause waits on _dayGate, cancel pending pace delays.
• Invert player ctor default: DecisionAttention = RunAlways (not UntilDecision). “To horizon” remains PauseMode.Never / attention ignored until complete.
• Avalonia: replace/augment transport strip with Attention control (3-way) + Speed slider; keep Step/Continue for HardPause and one-shot Step.

Primary files: `CampaignRunner.cs`, `PlayerControlState.cs`, `CaptainBridgeService.cs`, `MainWindow.cs`, session DTOs in Novolis.Game.Session.

2. You-are-here map marker (incl. transit)

• Extend star-map model with an explicit ship pose, separate from SelectedId:
• Docked: marker on CurrentSystemId.
• In transit: interpolate along the current corridor using shipment progress (LegIndex, SegmentHoursRemaining / LegHoursTotal when available; else sit on CurrentHubId).
• `StarMapControl`: add ShipWorldX/Y (or ShipMarker point) + distinct draw (not selection yellow). Selection stays for travel/job target.
• `CaptainBridgeModel`: compute pose each capture; always refresh marker (not only when a hub is selected). Detail pane can still say HERE when selected hub matches dock; marker is authoritative for “we are”.

3. Per-actor intent stack + state machine

Scope for this pass: Calypso/James as the first ICaptainActor (HTTP + UI). Scaffold the interface so later AI captains can plug in; do not rewrite all economy firm agents yet.

stateDiagram-v2
  [*] --> Idle
  Idle --> Running: Push intent or compound
  Running --> Running: Step succeeds / requeue substep
  Running --> Blocked: Needs input or resource
  Blocked --> Running: Player/agent supplies next order
  Running --> Idle: Stack empty and hull idle
  Blocked --> Idle: Cancel stack

• Introduce `CaptainIntentStack` on top of (or replacing the public face of) PlayerOrderQueue:
• Ordered stack of intents (atomic PlayerOrder or compound recipes).
• Compound examples: PrepareAndDepart → ensure premium (if required) → elective overhaul (if flagged) → bunker → lift cargo → PlanShipment/DepartManifest; Reposition → bunker → PlanReposition.
• Each step reports status (Pending, Active, WaitingFuel, WaitingCargo, Failed, Done) into snapshot / bridge panel.
• Drain policy: still one atomic order per agent tick, but the SM owns expansion/requeue instead of ad-hoc logic only inside DepartManifest/TravelTo.
• UI + session: “stack” strip listing pending steps; enqueue without requiring HardPause; Cancel stack clears queue + aborts compound. Bridge buttons push onto the stack (multi-decision) rather than only firing one-shot and hoping Continue.
• Move bunker/lift gates from opaque requeues into named stack steps (fixes false “depart ok” UX by surfacing WaitingCargo / WaitingFuel).
• NeedsPlayerDecision() under HardPause/SoftSlow: true when hull idle and (stack empty needing a new plan, or stack Blocked needing input)—not merely “orders count == 0”.

Primary files: new types under Universe/Player/ (e.g. CaptainIntentStack.cs, CaptainIntentKind.cs), `PlayerTrampAgent.cs`, `CaptainActions.cs`, bridge snapshot/actions, MainWindow stack panel.

4. Hygiene while touching this path

• Remove session #region agent log / debug-cc3657 instrumentation from `PlayerTrampAgent` / `StubPhases` once this work ships (or as first commit of the implementation branch).
• Keep cargo-lift-before-PlanShipment and dock-stock room gating (already proven); express them as stack steps.

Acceptance

• Default launch: time runs at selected speed with no decision hard-pause; slider moves from crawl (~1 real min / game hour) to max CPU (~day per pulse).
• Attention SoftSlow / HardPause behave as above; Step 1d still works.
• Star map always shows ship marker docked or along current leg.
• Captain can queue e.g. premium → fuel/lift → depart as a visible stack; HTTP agent can push the same sequence; progress/failure visible without false Ok.