fix(progression): an imported session is graded on all of its sets

[kurktchiev]

After a Hevy or CSV import, the first weight the app suggests for a lift can come from the light early sets. A session logged as 40/50/60/70/80 × 5 reads as a 60 kg session when the routine asks for 3 sets today, and the next prescription is 62.5.

#233 grades only the sets the plan asked for, and readSession takes that count from (entry && entry.target) || fallback. The importers write no target, and neither did anything logged before v1.2.2. For those entries the count comes from the exercise's current config, which sessionsFor passes in as the fallback, so the slice keeps the first three rows of a session that was never planned as three sets.

readSession now slices only when the entry carries its own target. An entry without one is graded on every work set, which is how it was read before #233. enough, the goal and the mode still come from the fallback, as before. api/coach/core/payload.js has its own copy of readSession and gets the same one-line rule, so the Coach's stall count keeps agreeing with the app.

Tests:

• progression.extra-sets.test.js: The ramp above reads 80 kg, and nextPrescription after it is 82.5. On the old progression.js they read 60 and 62.5. The file's 8 tests pass.
• payload-parity.test.js: An imported session whose fourth set falls short is a miss on both sides. It fails with either copy reverted. 4 pass.
• Also run: progression.test.js, assisted, exercise-history, progression-copy, session-start, coach-parity, import-hevy, import-csv, RoutineEdit.progression and Workout.noprog (410 tests across 11 files with the extra-sets file), Workout.test.jsx (119), mcp/test/tools.test.js (63), and api payload, payload-bounds, cohort, review-repair, jobs and prompts (76). All pass.

Overlaps:

• Per side on timed holds, rebased on v1.3.9 with the progression fix #322 edits the planned line just above this one and merges cleanly with it. This change only decides whether to slice, so Per side on timed holds, rebased on v1.3.9 with the progression fix #322's doubled count for a timed per-side hold still applies to an entry that carries its own target.
• Percentage (wave) progression + SetBlock refactor #168 and feat: add triple progression #181 rewrite readSession and already conflict with main inside it. This change falls inside those same conflicts, so whichever lands second has to keep the entry's own target as the only thing that slices. Percentage (wave) progression + SetBlock refactor #168 also conflicts with main in payload.js's copy.
• Add distance logging mode (metres stored; feet for imperial) #177 adds a distance branch to readSession and Fix: Double progression: grade against the top of the range, not the session's aim #301 changes nextPrescription. Neither gains a conflict from this beyond the ones it already has with main.
• feat(engine): generic prescription engine v2, smart warm-ups and v1→v2 migration #312 moves the frontend's progression into its engine and deletes progression.extra-sets.test.js, so it conflicts with this. It keeps payload.js's slice as it is on main, which would need the same rule.

🤖 Generated with Claude Code

[DuarteSantos8]

Merged and released in v1.3.10, thanks @kurktchiev!

https://github.com/DuarteSantos8/openGym/releases/tag/v1.3.10