Improve Windows debugger launcher

[fgiancane8]

While testing a trivial Rust application on my Windows machine, I discovered that rust-windbg.cmd could only launch windbg (legacy Windows debugger), could not detect windbgx, the more modern Windows debugger that fully supports Rust.
Prefer windbgx when available, fall back to windbg if the former is not available, propagate the debugger exit status, and report a clear error when neither debugger is found.

Executed tests:
[x] windbgx available -> launched correctly
[x] windbgx not available, windbg available -> legacy windbg called
[x] neither available -> error message printed

When a debugger is found and launched successfully, the script returns its exit code (0 normally). When neither debugger is found, PowerShell reports $LASTEXITCODE as 1.

LLM disclosure: I wrote the commit by hand after finding the issue myself. I used an LLM to review the commit before submitting.

[rustbot]

Thanks for the pull request, and welcome! The Rust Project has assigned @Mark-Simulacrum (or someone else) to review your changes, you should hear from them (or someone else) within the next two weeks.

Please see the contribution instructions and our LLM policy for more information.

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: @Mark-Simulacrum

[ChrisDenton]

Is the cmd script actually useful now that natvis files are embedded in the pdbs?

[fgiancane8]

Good catch. Probably it is now automatically managed by WinDbgX configuration but I haven't checked.
Are you suggesting there's no need to explicitly load natvis files any more or just recommending not to start debugging through the command line script at all?

[ChrisDenton]

I think the cmd scripts should be considered deprecated as they're now redundant. We could also consider removing them but if we do that we should emit a deprecation warning first.

[fgiancane8]

I think the cmd scripts should be considered deprecated as they're now redundant. We could also consider removing them but if we do that we should emit a deprecation warning first.

Thanks for the inputs. Should I update this pull request with a deprecation warning/message first, or shall we just ignore it here?

[Mark-Simulacrum]

r? ChrisDenton

cc @Walnut356