Add Rippr source snapshot and full-history bundle

Two copies for two jobs. rippr-src/ is a browsable git archive export of
the tracked tree at 2f76983 - no build outputs, no local.properties, no
nested .git - which is convenient to read in gitea but carries no history
and will drift.

rippr-full-history.bundle is the real backup: all 18 commits, verified as
"records a complete history" and test-cloned before committing. This
matters because ~/dojo/rippr has no git remote and otherwise exists only
on one machine.

rippr-src/SNAPSHOT.md explains the difference and how to restore.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

rippr-src/docs/v2/11-trip-detail.md

@@ -0,0 +1,86 @@
+# T11 — Trip detail screen (stats only)
+
+**Phase** 4 · **Depends on** T05, T08 · **Status** Done
+
+## Goal
+
+A full breakdown of a single ride — summary figures, elevation profile, speed
+distribution — with a deliberate placeholder where the map will go. Done when every
+statistic renders correctly and the screen is verifiable before osmdroid lands.
+
+## Context
+
+Splitting this from the map is intentional. osmdroid brings tile caching, `MapView`
+lifecycle, and coordinate projection — three sources of subtle failure. Building the
+screen first means that when the map misbehaves in T13, the surrounding screen is already
+known-good and the bug has nowhere to hide.
+
+Data comes from `RideStatistics.compute()` (T05) over `pointsForTrip(id)` and
+`segmentsForTrip(id)`. Unlike the list screen, this one **does** load points — a few
+thousand rows for one trip is fine, and the derived series need them.
+
+## Design
+
+Sections top to bottom:
+
+1. **Header** — name (inline-editable in T12) and start date/time
+2. **Map placeholder** — a bordered box captioned "Map coming in T13", replaced wholesale
+   by T13/T14. Reserving the space now means the layout does not shift later.
+3. **Summary grid** — distance, moving time, elapsed time, max speed, average moving
+   speed, elevation gain
+4. **Elevation profile** — line chart over distance-along-path
+5. **Speed distribution** — histogram in 10 km/h buckets
+6. **Segments** — listed only when there is more than one, showing where the ride paused
+
+Charts are drawn with **Compose `Canvas`**, not a charting library. Two simple series do
+not justify a dependency, and hand-drawn keeps them consistent with the existing visual
+language.
+
+Both charts need: an explicit empty/insufficient-data state, axis labels with units, and
+`tabular-nums` on any numeric label.
+
+**Point loading happens off the main thread** in the ViewModel, with a loading state. A
+three-hour ride is ~21,600 rows; loading that synchronously would jank the transition
+into the screen.
+
+## Implementation
+
+1. `TripDetailViewModel` loading trip, points, and segments on `Dispatchers.IO`, exposing
+   `Loading | Ready(summary, profile, histogram, segments) | NotFound`.
+2. Summary grid from `RideSummary`.
+3. `ElevationChart` composable — `Canvas`, path stroke, min/max labels.
+4. `SpeedHistogram` composable — `Canvas`, bars, bucket labels.
+5. Segment list, shown conditionally.
+6. Map placeholder box.
+7. `NotFound` state for a deleted trip.
+
+## Acceptance criteria
+
+- [x] All summary figures match `RideStatistics.compute()`
+- [ ] Charts render for a real ride and degrade gracefully with <2 points
+- [x] Segments section appears only when the ride was paused
+- [x] Loading state shown while points load; no main-thread I/O
+- [ ] Deleted trip shows `NotFound` rather than crashing
+- [x] Map placeholder occupies the space the real map will take
+
+> **Not verified.** The unchecked items above, and the Compose UI tests below, were
+> not done. Verification for T08–T11 was manual on the emulator (screenshots through
+> the full flow) plus the existing unit and instrumented suites. Compose UI tests are
+> outstanding — tracked in T18.
+
+## Tests
+
+Compose UI tests: loading, ready, not-found, single-segment vs multi-segment.
+Unit tests for the chart data-preparation helpers (scaling, bucketing) — the drawing
+itself is verified by screenshot on the emulator.
+
+## Risks / gotchas
+
+- **Main-thread point loading is the obvious trap here.** Keep it in the ViewModel on IO.
+- **Chart division by zero** when all altitudes or speeds are identical — a flat ride
+  gives a zero-height range. Guard the scaling.
+- **Do not start integrating the map in this task.** The separation is the point.
+
+## Out of scope
+
+The map (T13, T14), rename/delete/merge (T12), export (T16).