Replace the avatar's portrait

One image (jpg, png or webp, <= 10 MB) as the file part. A good portrait is front-facing, head and shoulders with some margin around the head (a face that fills the frame cannot be animated), the face at least about a third of the frame, in even light, with the lips slightly parted as if about to speak: a v2 avatar animates its first moments from the portrait itself, so a closed mouth usually makes the first second of speech fail (the lips stay still). The build refuses a clearly bad portrait at the appearance stage with a portrait_* reason in userFeedback (no face, face too small, lips closed). Or, on an instance that keeps uploads (where POST /v1/uploads answers), application/json: {"image": "<https URL | upload:<uploadId>>"}, the portrait by URL (downloaded like a JSON create's) or by a portrait upload from POST /v1/uploads, consumed when the replacement succeeds. Re-runs only the build stages that depend on this part; poll GET /v1/avatars/{avatarId} for progress. Requires avatars:write. A draft is given its portrait this way; on v2 it captures the first candidate once the voice is there too.