result: Handle testbed setup failures, treating as 'BAD'
If a test log lacks a '@@@@@@@@@@@@@@@@@@@@' marker, consider it to be
a failure during the testbed setup. Set an error message in this case,
and give it a single subtest set to a 'BAD' status.
result: Return triggers even for failed testbed setups
The log files use "@@@@@@@@@@@@@@@@@@@@" as a marker to separate the
different stages. For tests that fail during the initial setup, they
will only have a single section and thus won't have the separator.
Since the triggers are provided in the first section, we can return the
valid triggers for the test even though the run terminated early.
Commit f4147123 added an extraneous extra call to
get_autopkgtest_results(). This is a slow code path since it's
processing remote log files so the extra call can be a notable
performance impact.
packaging: Drop deprecated setup.py in favor of 'build' pkg system
The packaging has already partly transitioned to using the build
packaging system with a pyproject.toml and a Makefile. Complete the
transition by dropping setup.py and any reference to it in directions.
Add functionality to filter test results by package
Specifying the --package command line option has filtered the list of
triggers for a PPA, but previously it did not filter the test results.
For PPAs involving a variety of packages this can produce an excessive
listing.
This commit expands support for --package to apply not only to results
as well, but also the running and waiting lists.
io: Quell error messages for non-existing results and jobs
Suppress the "Could not retrieve data..." error messages printed by `ppa
tests`. To locate results, the command is scanning expected URLs, and
their absence simply means they're not available yet.
