Revisit parsing for `bucket-schema` commands

When the `bucket-schema` commands were implemented, I pushed to have all the logic for parsing schemas from-file-or-stdin live under the `clients/` package. Since then, I've changed my mind and started putting that sort of logic under `cmd/`. It would be nice to refactor the `bucket-schema` code to be consistent with the new pattern.

Start by locating the bucket-schema command parsing under the clients/ package and compare it with the newer parsing pattern under cmd/. Trace all from-file-or-stdin schema handling and identify the files that need to move or change; the refactor is done when bucket-schema follows the cmd/ pattern consistently.

Written by the indexing model from the issue text.