fix: guard the workout upload path against concurrent runs

[Anexus5919]

📌 Related Issue

Fixes #875

---

📝 Description

syncInProgress only wrapped the online-event handler, so a sync reached through fullSyncWorkouts (called unguarded on login in useWorkoutSync) could run concurrently with an online-triggered sync. Both call getUnsyncedWorkouts and read the same unsynced set before either marks them synced, and uploadWorkoutToFirestore uses addDoc (no idempotency), so each concurrent run creates a separate Firestore document for the same workout.

🔹 What has been changed?

• src/services/workoutSyncService.ts: moved the in-flight syncInProgress guard into syncWorkoutsToFirestore (the single upload choke point), so every entry point (login fullSyncWorkouts, the online handler, manual sync) shares it. Removed the now-redundant guard toggling from the online handler.

🔹 Why are these changes needed?

• The guard check and set run synchronously before the first await, so two concurrent callers can no longer both pass it and double-upload. This prevents duplicate workout documents in Firestore.

---

🛠️ Type of Change

• 🐛 Bug Fix

---

🧪 Testing

✅ Tests Performed

• Tested locally
• npx tsc --noEmit: changed file clean; npx eslint: 0 problems.
• Verified by reasoning about the synchronous guard: the if (syncInProgress) return / syncInProgress = true pair runs before any await, so concurrent callers are serialized to one upload run.

🔹 Note on automated tests

No committed unit test: this path needs both IndexedDB and a Firestore mock, and the repo has no fake-indexeddb harness. Verification is by tsc/eslint and the reasoning above.

🌐 Browsers Tested

Not applicable (sync logic).

---

📷 Screenshots / Demo (if applicable)

Not applicable.

---

📋 Checklist

• I have read the project's CONTRIBUTING guidelines
• My code follows the project style guidelines
• I have performed a self-review of my code
• I have tested my changes locally
• I have added/updated documentation where necessary (not applicable)
• My changes do not introduce new warnings or errors
• This PR is linked to an existing issue

---

💬 Additional Notes

Branched from latest upstream/main (5464425). A deeper follow-up (out of scope here) would give each workout a deterministic Firestore document id (e.g. setDoc keyed by crdtSessionId) so even a retry after a lost write acknowledgement cannot duplicate.

[vercel]

@Anexus5919 is attempting to deploy a commit to the somiljain2024-4175's projects Team on Vercel.

A member of the Team first needs to authorize it.

[Anexus5919]

@Somil450 @diksha78dev Kindly have a review on this pr. Thanks!