Explain tool decisions in generated output

[edsko]

When users use the high-level API, they might need ways to customize the generated output, sometimes perhaps even extensive customization. To make this easier, the generated output should

• be readable (pretty-printed)
• contain comments that explain decisions made by the tool (e.g., "rule such and such fired").

The latter especially will help users understand why the tool did what it did, and provide clues on how to modify that.