alsa: Add configurable delay after sample rate change

[wtp128pro]

Summary

• Adds a write_silence_locked() helper that writes zero-filled frames to the PCM device via snd_pcm_writei for a configurable duration
• Hooks into start_playback() immediately after snd_pcm_prepare() succeeds, so silence is written while the mutex is held and the pump thread is blocked — the only point where a direct write survives to hardware
• Detects rate changes (including cold start, where the previous rate is 0) in open_audio() and arms the silence duration if the setting is non-zero
• Exposes the setting as a spin button in the ALSA plugin preferences (0–10000 ms, step 10, default 0 = disabled)

Motivation

Some hardware DACs need a short settling period when the PCM sample rate changes. Without a delay, the first few milliseconds of audio after a rate transition are lost or corrupted. This is also observable on cold start when audacious first opens the audio device.

Test plan

• Set delay to 0 — verify no audible change in behaviour for same-rate tracks
• Set delay to 150 ms — verify silence is heard between tracks with different sample rates
• Set delay to 150 ms — verify silence is heard at the start of the first track after launch
• Verify the spin button appears in Preferences → Output → ALSA
• Verify the setting persists across restarts

🤖 Generated with Claude Code

[jlindgren90]

I don't like this. It is AI-generated, right?

From a quick scan:

• Doesn't follow modern C++ best practices (raw new/delete)
• Uses static variables with difficult-to-follow lifetimes
• Should have probably been an effect plugin to begin with

Sorry, I don't have time to review in more detail.