fix(voice): cap reconnect backoff at maximumConnectDelay - #586

doReconnect's timer is time.Duration(try) * delay instead of just delay. Since try never resets, the actual wait between retries grows linearly forever instead of capping at maximumConnectDelay (10s) — even though the comment above it says "Exponentially backoff up to a limit of 10s".

Confirmed against production logs: by try ~200+, retries were landing ~2000s+ apart (try * 10s), instead of holding steady at 10s.

Fix: use delay directly for the timer.