Tests resulting in the fatal-error state are counted as failures, but
are not currently reported in the text summary. This has already been
fixed for JSON output.
This fixes an issue with json report output that causes a crash due to
the incorrect object member being accessed for test-too-long. This also
changes treatment of fatal errors to match how they are displayed in the
tester log output and changes "fatal error" to "fatal-error" to better
match the naming of all the other failure types.
Change the subject to place the arch/bsp first and then a short
score card for the tests so the archive can be listed by subject and
you can see regressions.
The exe filter lets a BSP change the executable file to something
that can be downloaded to the target. For example U-Boot requires the image
format. The tester can now be configured to on-demand generate a specific
image for the target as the tester runs.
The conversion to Python3 added an eval() call which is wrong.
Fix the spelling in execute.
Fix labels in the tester gdb locking.
Check the debug-trace arguments.
Close#2642.
