Smarter cable trust signals and a few charging fixes

What's new

Cable trust signals. WhatCable now flags cables whose advertised e-marker data looks implausible. Phase 1 catches zeroed VID/PID and impossible PDO values, the safest of the heuristics. Wording stays hedged ("looks unusual," "common counterfeit pattern"), never "this cable is fake."
Cable reports include raw VDOs. The Markdown export now embeds the cable's raw Vendor Defined Objects so they can be inspected or quoted in bug reports without a separate JSON dump.

Don't blame the cable on USB-only ports. Front USB-C ports on the M4 Mac Mini hang off a plain xHCI controller and physically can't run USB-PD Discover Identity. WhatCable used to call those "basic cable", which wrongly blamed the cable for something the OS could never have read. Those ports now say "This port can't read cable details (USB-only port, no Power Delivery)" instead. JSON output gains a per-port pdCapable field. Thanks to @jlbyrne-76 for the report (#50).
Don't fabricate per-port charging wattage. With two chargers attached at once (e.g. an 87W adapter on one port, a 30W power bank on the other), the second port's diagnostic could borrow the first port's system-wide adapter wattage and claim "Charging well at 87W" on the 30W port. The diagnostic now refuses to substitute system-wide values when the per-port source has nothing real to report. Thanks to @joeshaw for the report (#46).
Don't show empty ports as connected. The popover counted any cached USB-PD power source as proof a port was live. Because that cache outlives the physical connection, disconnected ports could light up as "connected and charging" with stale wattage. Power sources now have to be backed by an active port-level connection before they count. Thanks to @bigskookum for the report (#47).