The list of changed files is pulled from hg.mo by revision hash, so all that really matters when testing locally is that comm_head_revision in your parameters.yml file points to a Thunderbird code push or a suite-only push.
You'll need a parameters.yml file from a push to comm-central that was a "suite-only" push. This is the most recent one:
The Decision task artifacts will have the parameters.yml file, download and save someplace.
Full testing will require four runs of mach taskgraph optimized.
Run 1: Thunderbird code push with current code
time ./mach taskgraph optimized -p project=comm-central --root comm/taskcluster/ci -o /tmp/run1.txt
Run 2: Suite-only code push with current code
time ./mach taskgraph optimized -p /tmp/suiteonly.yml --root comm/taskcluster/ci -o /tmp/run2.txt
run1.txt should contain ~170 jobs that would run (give or take depending on if there are toolchain jobs and such to run).
run2.txt should have two source-test-* jobs.
Apply the patch from this bug locally and repeat the above as run3 and run4.
Run 3: Thunderbird code push with patch
time ./mach taskgraph optimized -p project=comm-central --root comm/taskcluster/ci -o /tmp/run3.txt
Run 4: Suite-only code push with patch
time ./mach taskgraph optimized -p /tmp/suiteonly.yml --root comm/taskcluster/ci -o /tmp/run4.txt
run3.txt should have the same jobs as run1.txt (sort the files before comparing), and run4.txt should match run2.txt.
The Nightly build taskgraphs should be checked as well. That's going to be more challenging. Grab the parameters.yml from the latest Nightly. It's decision task is cron(Nd). Run mach taskgraph optimized with that parameters file without the patch then with the patch. To test a suite-only "nightly" push you'll need to make a copy of that nightly parameters file and change comm_head_revision to the one from your suite-only paramters from above. Then run mach taskgraph optimized without and with the patch.
Nightly builds are different. If the most recent push was a Suite-only push, the builds should still run. So there should not be any differences in the optimized job lists.