This PR has more changes to the diagnostics infrastructure.
The big one is centralizing the declarations of all the diagnostics messages (which involved changes at hundreds of call sites).
I added a few test cases for random error messages to try to confirm that output is still as expected. There were a few small changes in output that I let slide, because they seem to have been typos or inconsistencies in the original code.
A more complete set of test cases for error messages is still needed, as is a cleanup pass over the diagnostics to reduce the amount of redundancy.
This PR has more changes to the diagnostics infrastructure.
The big one is centralizing the declarations of all the diagnostics messages (which involved changes at hundreds of call sites).
I added a few test cases for random error messages to try to confirm that output is still as expected. There were a few small changes in output that I let slide, because they seem to have been typos or inconsistencies in the original code.
A more complete set of test cases for error messages is still needed, as is a cleanup pass over the diagnostics to reduce the amount of redundancy.