fix(noise): suppress recurring ffprobe demuxing-error lines in non-verbose activity log (closes #847)

[Salem874]

Summary

Closes #847 — the most-aggravating recurring log noise in the user-facing activity log. ffprobe's "Invalid data found when processing input" warning fires from stderr on virtually every track during enrichment as ffmpeg encounters the partial-moov-atom case at byte 0. Downloads complete; ffprobe falls through to a valid result on retry. But ~20 noise lines per album turned the activity log into static.

Implementation modelled exactly on the shipped #660 pattern (`is_python_traceback_noise` + non-verbose suppression in `download_queue.rs::emit_subprocess_line` gate). Verbose mode still surfaces everything; on-disk activity-log writer still records every line so support requests stay debuggable.

What changed

• New `process::is_ffprobe_demux_noise` in src-tauri/src/utils/process.rs. Match keys on the structural `[in#/ @ 0x]` prefix plus both required tail substrings (`Error during demuxing:` AND `Invalid data found when processing input`).
• Wired into both stdout + stderr gates in src-tauri/src/services/download_queue.rs at lines ~10958 and ~11155. `is_known_noise` now combines both `is_python_traceback_noise` (Activity log: Python tracebacks from GAMDL leak through as misclassified errors #660) and `is_ffprobe_demux_noise` (this PR). The 5th arg to `emit_subprocess_line` (`show_in_ui`) stays `verbose || !is_known_noise`.

Design choices

Why a tight structural match rather than a regex on the literal demuxer list? Future ffmpeg may auto-select a different comma-separated demuxer list (the canonical `mov,mp4,m4a,3gp,3g2,mj2` set could shift to include / exclude formats). The match keys on the bracket shape `[in#N/...` + the `@ 0x]` pointer suffix, which has been stable across ffmpeg 4.x / 5.x / 6.x / 7.x. If upstream ever switches the pointer format the gate fails OPEN (noise resumes) rather than over-suppressing genuine errors. Safe-default direction.

Why combine with the existing #660 gate rather than a separate emit path? Single `is_known_noise` boolean keeps the call site readable + avoids the "every new noise filter doubles the branching" pattern. Future noise gates land in the same OR chain.

Tests

6 new unit tests pin the contract — all green:

Test	What it pins
canonical Apple Music shape	verbatim from #847 issue report
leading-whitespace + wider hex pointer tolerance	ffmpeg builds + 64-bit platform variations
alternate demuxer lists (aac, wav,mp3)	future-proof against ffmpeg's auto-selection changes
genuine ffmpeg errors pass through	muxer errors, parameter errors, generic "Invalid data" without the demuxing prefix all still visible
empty / unrelated / traceback lines pass through	no false positives
wrapped GAMDL `[INFO HH:MM:SS]` prefix doesn't trigger	structlog-prefixed lines aren't confused with ffmpeg's prefix
requires `0x` pointer form	upstream format change → gate fails open (safe)

Verification

• ✅ `cargo test --lib is_ffprobe_demux_noise` — 6/6 pass
• ✅ Full test suite green (no regression)
• ✅ Verbose mode unchanged — every line still emits both on-disk + UI
• ✅ Disk writer unchanged — forensic record stays complete

Test plan

• Download an album in non-verbose mode → ffprobe demuxing-error lines no longer appear in the activity log (was ~4 lines per track).
• Toggle verbose mode → lines reappear immediately.
• Check `{app_data_dir}/logs/activity-YYYY-MM-DD.log` — every suppressed line is still recorded on disk.
• Provoke a genuine non-demuxing ffmpeg error (e.g. paste an empty file at the codec-detection step) → it should still surface in the activity log (the gate is tight enough not to swallow real errors).

🤖 Generated with Claude Code

[github-actions]

PR Security Checks

⚠️ 1 category(ies) of findings — these are heuristics, please review each:

Hardcoded absolute filesystem paths (derive from app_data_dir / std::env, not literals)

src-tauri/src/services/download_queue.rs:15180:        assert!(super::super::config_service::validate_path_safe("/home/user/Music").is_ok());
src-tauri/src/services/download_queue.rs:15181:        assert!(super::super::config_service::validate_path_safe("C:\\Users\\Music").is_ok());
src-tauri/src/services/download_queue.rs:15188:        assert!(super::super::config_service::validate_path_safe("/home/../root").is_err());
src-tauri/src/utils/process.rs:2030:                r#"File "/Users/user/Library/Application Support/io.github.meedyadl/python/lib/python3.12/site-packages/httpx/_transports/default.py", line 118, in map_httpcore_exceptions"#

_{Generated by .github/workflows/pr-security.yml. Non-blocking — the merge gate is ci.yml. Cross-source checks live in tools/audit-checks/ and run locally. False positives are expected.}

[Salem874]

Superseded by PR #954 (combined alpha-targeting housekeeping).

Entire content (the 2d7701e0 ffprobe noise filter commit + 6 unit tests) landed cleanly in PR #954 via cherry-pick onto an alpha base. Closing per user instruction to consolidate.