Emitters need concrete syntax ordering from the TypeSpec file to be available so they can maintain that order in output files.

### Discussed in https://github.com/microsoft/typespec/discussions/3261

For our workload, grouping and ordering of input is important, whether in a single file or spread across multiple imported files.

As we're in the middle of a long-term migration, being able to diff manually created output against typespec-generated-output and have a rational experience is key to building confidence with our customers.

Today we're not able to replicate input ordering in output, meaning every diff is an adventure in trying to convince yourself that semantically meaningful changes haven't crept in in the transition to TypeSpec.

This is holding up our ability to migrate at scale, as the manual verification tax is so high.