Align get_smry and load_smry

[berland]

For a ScratchEnsemble, get_smry() and load_smry() may differ in dates returned - get_smry() first builds a list of dates to obtain data for, and then asks every realization for data at these dates, while load_smry() asks each realization independently. This will only happen if the realizations summary data have different end dates.

It is not given that this difference is a bug and not a feature, so perhaps it should only be documented.

[berland]

If behaviour should change, get_smry() should be changed to mimic load_smry().

[berland]

get_smry() can change behaviour towards load_smry() for a 2.0 release of fmu-ensemble. Bumping major revision because of this change in API.

[berland]

See also #94

[berland]

See also #180

[berland]

Solved in #182

[berland]

From proposed git commit:

Align load/get_smry, real-specific time index

ensemble.get_smr() will no longer ensure that the dates returned
are identical in all realization, realizations with shorter timespans
will not be extrapolated as they were before.

This also affects get_smry_stats(), for which realizations
which failed prematurely earlier had zero rates, affecting statistical
profiles. Now these realization with truncated time-series will only
affect the statistical profiles in the time-range they have data for.