Detect when multiple commands have the same output

Asset Cooker should mark these commands as errors and refuse to cook them.

One step further would be to validate that commands using outputs of other commands have a lower prio (ie. reliably cook *after*).

The issue names no files, tests, or entry points. Start by locating command output handling and the cooking dependency-order logic; done means duplicate outputs are reported as errors and cooking is refused, with dependent commands validated to run after the commands producing their inputs.

Written by the indexing model from the issue text.