"Clock latched" reads no forever on WHOOP 5.0/MG

[jbergler]

Problem

The Devices/Test Centre "Clock latched" readout (ConnectionReadout.clockLatchedLabel) reads "no (waiting for the strap clock)" on every WHOOP 5.0/MG connection, regardless of whether the strap's clock is actually set — not just on a fresh strap with no banked history.

Root cause

GET_CLOCK's COMMAND_RESPONSE is sent to WHOOP 5/MG straps (as part of the #78 fix that made history offload work) and the strap does reply, but the reply is silently dropped end-to-end:

• decodeWhoop5CommandResponse (Packages/WhoopProtocol/Sources/WhoopProtocol/Interpreter.swift) only handles GET_BATTERY_LEVEL, GET_DATA_RANGE, and GET_HELLO — there's no case for GET_CLOCK (cmd 11), so parsed["clock"] never gets set for a 5/MG frame.
• FrameRouter.swift's COMMAND_RESPONSE handling gates its per-command branches on family == .whoop4 — nothing routes a 5/MG GET_CLOCK reply anywhere.
• The documented fallback signal (strapNewestUnix, from GET_DATA_RANGE, added in on DEVICES, the strap says Clock latched: no #261) is also dead for 5/MG: GET_DATA_RANGE (cmd 34) is never sent to a 5/MG strap at all — the only call site is inside the WHOOP4-only connect-handshake block.

So both inputs to clockLatchedLabel are structurally nil for the whole 5/MG family.

Plan

1. Add a GET_CLOCK debug probe (mirroring the #592/#690 extended-battery/body-location probe pattern) to capture the raw WHOOP 5.0 GET_CLOCK COMMAND_RESPONSE payload from real hardware.
2. Use the captured bytes to write a real decodeWhoop5CommandResponse case for GET_CLOCK, wire it through FrameRouter, and feed a real signal into clockLatchedLabel for 5/MG.
3. Android: deferred to a follow-up once the Swift/macOS side is confirmed on hardware.

Note: GET_DATA_RANGE never being sent for 5/MG at all is a separate, likely more impactful gap — tracked separately, out of scope here.

[jbergler]

I'll work on this, just wanted an issue number to reference for comments.