[GLinnik21]

When Force Direct Play refuses a file because of its audio, the read-out now names the codec ("…this audio (DTS) can't play without conversion") instead of "this audio format". Field reports like the one in discussion #695 and the C1 refusal on Reddit could not tell us which codec hit the wall.

• ForcedFailure::Audio { codec } and AudioNeedsConversion { codec } carry friendly_codec's name for the track the viewer was going to play (session pick → remembered/selected → language → file default → first track → Media-level acodec). An unknown codec keeps the old generic sentence.
• Two new keys with {codec} in all 8 locales: widgets.verdict.forced_audio_codec, widgets.verdict.forced_audio_conversion_codec.
• Both refusal sites log playback_policy: Force Direct Play refused audio codec <raw> (<friendly>), so a log tells dca from dts. Telemetry payload unchanged (it is consent-previewed).
• PlayVerdict::text() returns Cow<str>; only the two codec sentences allocate, on the failure screen only. ForcedFailure is no longer Copy.

No layout or colour change: the read-out shows the verdict sentence where it already did.

Verification

• Red first: forced_audio_refusal_names_the_codec (dca → "DTS", truehd → "Dolby TrueHD") failed to compile on main (no codec field), then passed; plus forced_audio_refusal_without_a_codec_keeps_the_generic_sentence.
• make check on the committed tree: every test result: ok, 0 failed (largest: 1848 passed; 0 failed; 64 ignored). make check-nodef clean.
• Not done: a ui-sim capture of the read-out; the translations other than English are unreviewed (Belarusian reuses the catalogue's existing wording; Hungarian avoids a suffix on the codec name).

First PR of the DTS/lossless audio plan; the dca spelling fix stacks on this one.

🤖 Generated with Claude Code