Two MSVC-specific APIs missing from MinGW headers:

1. TCP_INFO_v0 and SIO_TCP_INFO (Windows 10 1703+ kernel RTT) are
   defined in MSVC's mstcpip.h but not MinGW's. Added manual
   struct/macro definitions guarded by #ifndef SIO_TCP_INFO.

2. _dupenv_s is MSVC-specific secure CRT. Replaced with standard
   std::getenv("APPDATA") for Thetis FFTW wisdom import path.

Co-Authored-By: JJ Boyd ~ boydsoftprez ~ KG4VCF <kg4vcf@gmail.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

…ethersdr#625)

The TunerApplet power and SWR gauges always showed 0 because meter data
comes from the TGXL's direct TCP connection (port 9010), not the radio's
VITA-49 meter stream. Three issues fixed:

1. TgxlConnection discarded all R responses (status poll replies) which
   contain fwd= and swr= fields. Now parses R response bodies as KV
   pairs and emits statusUpdated().

2. TunerModel had no fwd/swr fields. Now parses fwd (dBm→watts) and
   swr (return loss dB→ratio) from both stateUpdated and statusUpdated
   signals, emitting metersChanged() for the gauge.

3. Removed txMetersChanged→TunerApplet connection that was overwriting
   TGXL readings with exciter power values.

TGXL protocol: fwd is reported in dBm (watts = 10^(dBm/10)/1000),
swr is reported as negative return loss in dB
(rho = 10^(rl/20), SWR = (1+rho)/(1-rho)).

Fixes aethersdr#625

Co-Authored-By: JJ Boyd ~ boydsoftprez ~ KG4VCF <kg4vcf@gmail.com>
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

… ordering

macOS Core Audio claims to support 24kHz mic input via isFormatSupported()
but its internal resampler produces gravelly/distorted audio at that
non-standard rate. The previous code also had a macOS-only path that only
tried 48kHz and 24kHz stereo, failing entirely on the Mac built-in mic
(mono-only device).

Replace the platform-split negotiation with a unified loop that tries
stereo then mono across three rates. macOS prefers 48kHz first (letting
r8brain handle the clean 2:1 downsample to 24kHz), while Linux/Windows
prefer 24kHz first (avoiding resampling when possible).

Verified on-air: clean TX audio in both Opus and uncompressed modes
using Mac built-in mic (mono 48kHz capture → r8brain → 24kHz stereo).

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>