Implement per-job single retry of the failed tests

[asirko-soft]

Background: Flaky tests are a known source of instability in our validation workflows, especially in the REPL and Darwin jobs. These transient failures (such as network errors or timeouts due to a slow runner) cause entire jobs to fail for reasons unrelated to the code changes, forcing developers to manually intervene. This wastes time and consumes unnecessary compute resources.

Proposal: Implement a selective, automated retry mechanism. Instead of failing the entire job on the first error, the workflow will capture the names of failed tests and re-run only those specific tests. The job will be marked as successful if the retry passes and will only fail if the same tests fail a second time. Flakiness data will be captured as a build artifact for long-term analysis.

[asirko-soft]

Discovered how the tests are executing currently

suites (YAML Tests):

• Executor: scripts/tests/run_test_suite.py
• Mechanism: This script discovers YAML test files (typically in src/app/tests/suites), filters them based on arguments (tags like PURPOSEFUL_FAILURE or explicit targets), and executes them using a runner (e.g., chip-tool or matter-repl).
• Workflow Usage: In test_suites_linux, it's called via scripts/tests/run_test_suite.py with arguments like --runner chip_tool_python and --target ....

matter_testing_infrastructure (Python Tests):

• Executor: src/python_testing/execute_python_tests.py
• Mechanism: This script scans a directory (defaulting to src/python_testing) for Python test scripts (matching *.py), filters them using glob/regex patterns, and executes each script as a subprocess.
• Workflow Usage: In repl_tests_linux_run, it's called via src/python_testing/execute_python_tests.py with --search-directory src/python_testing.

To implement "Retry on Failure" workflow, I need to modify the test runners to support exporting and importing failure lists.