These tests are only testing the resolution, but were running full apps (which are ~20s) for no real reason.

We ran this test 3 times to check after 0, 1, 2 breakpoints - however each test effectively included the previous. Running just the last one gives the same coverage but removes ~40s out of the ~60s to run the three.

This test was designed to catch the difference between original source, and source after a reload. Two reloads shouldn't be any different to one (but makes the test a fair bit longer - especially for web).

…g to launch

See #2147.

Revisit this once there's a response or fix to microsoft/vscode#86403.

Trying to track down intermittent issue where on Windows the sandbox doesn't seem to be intercepting the calls to the dart.getPackages command in pub_get_runs_automatically_when_pubspec_is_saved.