Add persistent driver training history and progression

docs/javascript-architecture.md

@@ -399,16 +399,21 @@ Type:
 
 Owns:
 - training-state normalization
-- valid driver-lap aggregation across `resultsBySession`
-- lifetime/session/best-lap/consistency calculations
+- durable ledger synchronization from `resultsBySession`
+- valid driver-lap aggregation across the training ledger
+- lifetime/session/day/best-lap/average/consistency calculations
+- daily summaries, controlled-lap streak, and recent personal-best gain
 - rule-based practice recommendation selection
-- milestone progress calculation
+- automatic and custom milestone progress calculation
 
 Rules:
 - only passings with a matching `driverId` are included
 - invalid laps and lap times at or below `500 ms` are excluded
 - anonymous Open Practice transponders are excluded until registered as drivers
 - consistency is the P10-to-P90 spread of the latest 30 valid laps divided by their mean
+- every save replaces ledger rows only for session IDs that still exist in `resultsBySession`
+- ledger rows for a deleted session are retained and become historical records
+- `sourceKey` deduplicates repeated saves of the same session passing
 
 ### `src/training_views.js`
 Type:
@@ -417,6 +422,7 @@ Type:
 Owns:
 - driver selector
 - training stat cards and recommendation display
+- automatic next-level card and daily practice history
 - milestone/reward create, claim, and delete actions
 - mobile-responsive training markup
 
@@ -1435,7 +1441,8 @@ This bucket is the most dynamic part of state and is mutated heavily during live
 
 ```js
 {
-  goals: TrainingGoal[]
+  goals: TrainingGoal[],
+  ledger: TrainingLap[]
 }
 ```
 
@@ -1453,20 +1460,40 @@ This bucket is the most dynamic part of state and is mutated heavily during live
 }
 ```
 
+`TrainingLap`:
+
+```js
+{
+  id: string,
+  sourceKey: string, // sessionId + passing timestamp + competitor + passing index
+  sessionId: string,
+  sessionName: string,
+  eventId: string,
+  eventName: string,
+  driverId: string,
+  timestamp: number,
+  lapMs: number,
+  recordedAt: string
+}
+```
+
 Normalized by:
 - `normalizeTrainingState()` wrapper in `src/app.js`
 - `normalizeTrainingState()` implementation in `src/training_logic.js`
+- `syncTrainingLedger()` wrapper in `src/app.js`
+- `syncTrainingLedger()` implementation in `src/training_logic.js`
 
 Persisted through:
 - `buildPersistableState()` in `src/app.js`
 - `applyPersistedStateHelper()` in `src/runtime_services.js`
 - existing full-state `POST /api/state` backend path
 
-Derived statistics are intentionally not stored. They are recalculated from `resultsBySession` so corrections and invalidated laps are reflected immediately.
+Derived statistics are intentionally not stored. They are recalculated from `state.training.ledger`. Before persistence, `saveState()` synchronizes every session currently present in `resultsBySession`: existing rows for those session IDs are replaced from current valid passings, so judging corrections are reflected. Session IDs absent from `resultsBySession` are not removed from the ledger, preserving lifetime totals after archival or deletion.
 
-Known lifecycle constraint:
-- deleting historical sessions removes their source passings and therefore lowers calculated lifetime totals
-- a future immutable training ledger should be introduced before session archival/deletion is used routinely
+Migration behavior:
+- older state without `training.ledger` normalizes to an empty ledger
+- the next `saveState()` imports valid registered-driver laps from all current session results
+- invalid laps, anonymous transponders, and laps at or below `500 ms` are excluded
 
 ## 15.9 `Settings`