[Breaking change] Include starting time (0) in `get_algorithm_time`

[ArneBouillon]

It seems useful if the entries of get_algorithm_time correspond to the ensembles in get_u/get_g. This would allow, for instance,

for (ensemble, t) in zip(get_u(ekp), get_algorithm_time(ekp))
    # ...
end

Currently, the initial timepoint (0) is not included in get_algorithm_time, while the initial ensemble is in get_u.

Changing that might break people's code, however (including mine, now that I've worked around this).