korad-kaxxxxp: Add KA3305P support, handle multiple outputs

[minuteman3]

Adds support for the Korad KA3305P and its RND 320-KA3305P rebadge — a dual-output variant of the KA3005P family — and teaches the driver to handle more than one output.

Model support

• Detected natively (KORAD KA3305P V7.1), no force_detect workaround.
• Ranges per the manual: 0–30 V / 0–5 A, 10 mV / 1 mA resolution.

Multiple outputs

• Per-output values in dev_context; VSET/ISET/VOUT/IOUT now carry the output number instead of being hardcoded to output 1.
• Dual-output models expose V1/I1/V2/I2 plus one channel group per output, so per-output config can be addressed with e.g. sigrok-cli -g CH2.
• Single-output models are unchanged (V/I, previous behaviour).
• Output enable stays device-level: OUT<boolean> takes no channel on this family, so the outputs cannot be switched individually.

STATUS bit quirk

Measurements on real hardware contradict the documented STATUS? map, so this family reads OVP from bit 4 (gated behind KORAD_QUIRK_STATUS_OVP_BIT4):

bit	documented	driver used to read	actually
4	N/A	beep_enabled	OVP enable
5	N/A	ocp_enabled	OCP (unchanged)
6	Output	output	output (unchanged)
7	N/A	ovp_enabled	tracks output state, not OVP

As a result SR_CONF_OVER_VOLTAGE_PROTECTION_ENABLED reported true whenever the output was enabled and could never report false. The beeper has no status bit at all (toggling it changes no bit), so it is no longer derived from the OVP bit. Other models keep the previous bit 7 decoding.

Verification

Tested on a KA3305P (firmware V7.1) with a Brymen BM257s on CH2 as an independent witness:

• native detection: Korad KA3305P [S/N: 00018549] with 4 channels: V1 I1 V2 I2
• per-output routing with distinct values (CH1 0.5 A / 9.0 V vs CH2 0.1 A / 6.0 V); -g CH2 emits VSET2:…/ISET2:…, -g CH1 emits VSET1:…
• CH2 setpoint change independently confirmed by the DMM (6.00 → 7.50 → 6.00 V) while V1 stayed at 9.00 V
• ocp_enabled and ovp_enabled now round-trip correctly (set false → get false); the latter was previously impossible while the output was on
• bit map established from repeated OCP0/1, OVP0/1, BEEP0/1, TRACK0/1/2 sweeps
• make check passes; builds clean under -Wall -Wextra -Wmissing-prototypes

Caveats

• bit 4 may mean "OVP armed" or "OVP set mode"; these cannot be distinguished, as OVPSTE is write-only (no readback query) and is absent from the official manual, and no functional OVP trip was reproducible.
• OCP functional trip is untested (requires a load).
• The RND 320-KA3305P entry is inferred from documentation, not verified against hardware.