Change default PR-triggered test behavior

[leewujung]

Right now the PR-triggered tests are determined by the existence of the string [all tests ci] in the PR title. If this doesn't exist, only tests associated the module that was modified are run. This has caused quite a bit of problems since some code changes would break downstream cascading processing but would not be caught in this way.

echopype/.github/workflows/pr.yaml

Lines 72 to 79 in 3df9d83

	- name: Running all Tests
	if: contains(github.event.pull_request.title, '[all tests ci]')
	run: |
	pytest -vvv -rx --numprocesses=${{ env.NUM_WORKERS }} --max-worker-restart=3 --cov=echopype --cov-report=xml --log-cli-level=WARNING --disable-warnings
	- name: Running Tests
	if: ${{ !contains(github.event.pull_request.title, '[all tests ci]') }}
	run: |
	python .ci_helpers/run-test.py --pytest-args="--log-cli-level=WARNING,-vvv,-rx,--numprocesses=${{ env.NUM_WORKERS }},--max-worker-restart=3,--disable-warnings" --include-cov ${{ steps.files.outputs.added_modified_renamed }}

I am thinking of changing the default behavior to:

• running all tests whenever there is a code change
• if there are only docs changes, only the docs build are run

Or that we can make it such that:

• a label like "run CI" will trigger all test runs
• a label like "build docs" will trigger docs build
• otherwise nothing will be done

@oftfrfbf @ctuguinay : Any thoughts or suggestions?

Also, there was a label "Needs Complete Testing" that did not have any effect and was probably a relic from long time ago. I have removed it.