ctrace: Support default file name scheme changes in spec (for TB channel)

[jreineckearm]

Describe The Problem To Be Solved
The trace specification has evolved to support selection from multiple trace buffers in a system.
These changes have an impact on default file names (input: *.ctrace-run.yml / *.<channel>.raw, output: *.<channel>.csv) which need to be reflected in the ctrace decoder.

Related specification changes in PR Open-CMSIS-Pack/cmsis-toolbox#675

(Optional): Suggest A Solution

• Extend default trace output file naming scheme to TB, support name extension as specified.
  • If multiple trace: child nodes available in cbuild-run file, then use the first enabled. As per spec there currently must be only one list entry. But please gracefully ignore potential other entries in case we need to experiment with multiple trace sinks and hence multiple list nodes.
• TB naming scheme also applies to CSV output files.

Update, 04/09/2026: Discussion showed we may need to store more information in *.ctrace-run.yml before taking action. ctrace tool is not supposed to parse *.cbuild-run.yml directly which holds information on active trace sink selection.

[thorstendb-ARM]

Updated design direction

I discussed this with Jens yesterday. Instead of adding further filename heuristics, we intend to make *.ctrace-run.yml the authoritative source for raw trace discovery.

The proposed workflow is:

1. ctrace selects the trace-run configuration:

   • without --target: process all *.ctrace-run.yml files;
   • with --target <solution-set>: process only <solution-set>.ctrace-run.yml.

2. Each *.ctrace-run.yml contains a list describing the available raw trace inputs.

3. The listed entries are used to locate the corresponding raw trace files. The representation could contain either:

   • the complete name <solution-set>.<traceChannelName>.raw; or
   • only <traceChannelName>, with the filename derived by ctrace.

4. ctrace reads and converts only the raw trace files referenced by the selected *.ctrace-run.yml.

This avoids discovering trace inputs through an expanding set of filename heuristics. It also removes the need for ctrace to inspect *.cbuild-run.yml for this purpose.

Making *.ctrace-run.yml authoritative also improves validation and diagnostics between the user's trace configuration and the configuration actually resolved and applied to the target. Missing, unavailable, or mismatched trace channels and raw files can therefore be reported in the context of the original user configuration.

This is currently a working hypothesis. The exact ctrace-run.yml schema and filename resolution rules still need to be defined.