feat(audio): AutoEQ headphone correction with a native parametric EQ

[xppenTC]

What

AutoEQ headphone correction (like Poweramp / Wavelet), on top of a new native parametric EQ. Settings → Audio → Output & Loudness → AutoEQ.

How

• Data: the public AutoEQ database (MIT, ~8 800 results, every major lab). There is no official API: results/INDEX.md is downloaded on demand and searched locally. Only the chosen profile (ParametricEQ.txt) and, for the chart, its measurement CSV are downloaded. Everything is cached, so a chosen profile keeps working offline.
• Native EQ (ParametricEq.cpp):
  • up to 20 RBJ biquads (PK, low/high shelf, LP/HP) plus preamp;
  • one process-wide config, picked up lock-free by every instance;
  • filter state is kept per filter across edits, so moving a slider does not click.
• Where it runs:
  • first in DspProcessor::process, also while Bit-Perfect is on (the user explicitly asked for the correction);
  • in place on the routes that bypass the Float32 DSP: USB exclusive, internal direct output, Bit-Perfect AudioTrack. DirectPcmEq works in the source packing (16/24/32-bit int with TPDF dither for 16-bit, float). The route keeps its format, so AutoEQ + mixer bypass work together.
  • The signal path adds an "AutoEQ" row; badge and verdict show DIRECT + EQ (never bit-perfect).
• Targets:
  • default = the lab's own target, i.e. the published profile as-is;
  • Harman over-ear 2013/2015/2018, in-ear 2016/2017-1/2017-2/2019 (with and without bass), oratory1990 over-ear/in-ear, AutoEq in-ear, diffuse field ISO 11904-1 / GRAS KE