remove_solution does not remove timeseries with meta=0

[gidden]

This basically keeps all reported/post process timeseries in the clone/removed solution scenario. Thus if reporting is run again, stale values could be still included

[zikolach]

As far as I see removeSolution has nothing to do with timeseries. It removes variables and equations (see at.ac.iiasa.ixmp.objects.Scenario#removeSolution)

[danielhuppmann]

When implementing tests for this issue, this should also cover that clone(first_model_year=y) drops all timeseries data for ixmp.Scenario after (including) y and that the keep_solution=False is implemented as expected for message_ix.Scenario instances.