fix(analytics): normalize weight units (kg/lbs) in exercise stats - #227

What

All three exercise statistics routes plot/aggregate raw weight values without normalizing units, so a user who switches between kg and lb (or mixes them across sessions) gets a misleading trend line. A 50 lb set and a 50 kg set both read as "50", and a 110 lb set sits far below a 50 kg set on the same axis.

Bug found while auditing the analytics code — not covered by any existing issue/PR.

Root cause

Each route reads the weight straight out of the value array and never consults units[weightIndex]:

• weight-progression/route.ts: const weight = set.valuesInt[weightIndex]
• volume/route.ts: volume = reps × set.valuesInt[weightIndex]
• one-rep-max/route.ts: const weight = set.valuesInt[weightIndex]

The set stores its unit ("kg" / "lbs") at units[weightIndex], but it's ignored. The app already ships convertWeight() / WEIGHT_CONVERSION in src/shared/lib/weight-conversion.ts.

Fix

Normalize each weight to kg before aggregating, in all three routes:

import { convertWeight } from "@/shared/lib/weight-conversion";

const rawWeight = set.valuesInt[weightIndex];
const unit = set.units?.[weightIndex] === "lbs" ? "lbs" : "kg";
const weight = convertWeight(rawWeight, unit, "kg");

Now a 50 lb and a 50 kg set are no longer treated as equal, and mid-history unit switches don't make the line jump. Aggregates (per-session max weight, total volume, estimated 1RM) become unit-consistent.

Fixes #226.