BOLT-1217 PR #11955: refactor to reuse dischargeStrategy + fix usedDefault detection

BOLT-1217 PR #11955 rework based on reviewer feedback (2026-06-12 Slack thread). Two fixes: (1) Charge-side assumed-SoC constant fix. Add new config key smoothAssumedStartingSocForCharge (env DEVICE_SMOOTH_ASSUMED_STARTING_SOC_FOR_CHARGE), default 0.20. Charge calc must NOT reuse the discharge 0.85 — that produces 0W on charge-to-≤85% and trickle on charge-to-100%. Update calculateOptimalChargePower JSDoc to explain WHY the charge prior is opposite (over-shooting ceiling is clamped device-side, under-charging is the real risk → assume low starting SoC for big headroom). (2) Drop resolved.chargePowerRate / resolved.chargePowerPercent fork. Reuse dischargePowerRate / dischargePowerPercent end-to-end as positive magnitudes for both modes. Remove the charge-mode short-circuit in mappers.ts. Loud JSDoc on dischargePowerRate field re: positive magnitude, operatingMode disambiguates direction, consumer responsible for negation when mapping to signed power axis. Audit results 2026-06-12: all 5 charge-capable OEM adapters (EG4, FranklinWH, TGS, SolarEdge, Enphase) already branch on operatingMode before reading dischargePowerRate — Choice A is safe. Branch winston/bolt-1217-docs-discharge-power-interpretation. PR #11955.