Video-to-.tglyph animation conversion. Native-only (uses ffmpeg via
vectomancy-video, which can't target wasm32), so this crate is
intentionally not part of the wasm-facing dependency graph (see
topoglyph-docs/TODO.md 0.5.0 and the comment in the workspace
Cargo.toml).
A video is treated as nothing more than a sequence of images: each
decoded frame is run through the exact same
smoothing -> grid-clipping -> glyph-matching pipeline the still-image CLI
path uses (topoglyph_vectomancy::adapter, topoglyph_core::clipping,
topoglyph_core::matching), then the resulting [TextCanvas] sequence
is handed to [topoglyph_output::animation::TglyphAnimation] for
delta-encoding into the final .tglyph text document.