I believe it’s just a naming convention to group a set of related tasks. For example you might have
"test:ci": ...
"test:units": ....
"test:integration"...
In this case it is grouping a related set of test tasks.
It would be down to the package author to specify. You can split tasks out like described in the answer above and then have a ‘global’ test command which combines each of them e.g. test:ci && test:unit && test:integration enabling you to run them all at once or when individually when needed.
You can use npm-run-all (link) and use the command npm-run-all test:*, which would then find all scripts starting with the test: group.