murmur 0.3.3

Five commits since v0.3.2, all of them about what you actually hear. The radio
no longer opens on half a minute of silence — it says something about two
seconds after launch, and signs off on the way out instead of cutting the
sound. The pick stopped replaying the same shortlist: it knows what you already
know across every platform you have mounted, and keeps somewhere to go besides
memory. And the default female voice is no longer a copy of a copy of a copy —
it is cloned from a real recording.

One note on that last one: it changes the clip a voice is cloned from, not
any voice already pinned in your ~/.murmur/voice.json. A new listener picking
the female timbre during setup gets it automatically. If you are already
running murmur and want it, reopen the setup conversation (murmur --setup, or
/setup in the TUI) and say you want to change the voice — it will clone the
new clip under your own key.

What's Changed

• feat(music): stop handing the brain a playlist to replay [spec 14] by @wine-fall in #288
• feat(music): give the pick somewhere to go besides memory [spec 14] by @wine-fall in #289
• feat(music): let the pick know what the listener already knows [spec 14] by @wine-fall in #291
• feat(talk): open and sign off from stock lines [spec 04] by @wine-fall in #292
• feat(voice): clone the default female voice from a real recording [spec 03-03] by @wine-fall in #296
• chore(release): 0.3.3 by @wine-fall in #297

Full Changelog: v0.3.2...v0.3.3