Feature request for a method which allows one to subspace something which has a structure

Feature request for a cf.Field.group() method like subspace, but which allows one to subspace something which has a structure:

E.g., Subsetting data only for the extended winter (ONDJFM) from 2018-2020:

>>> ondjfm = a.subspace(T=cf.seasons(2, 10)[0] & cf.wi(cf.dt('2018-10-1'), cf.dt('2020-03-1')))
>>> print(ondjfm.coordinate("long_name=time").dtarray)
[cftime.DatetimeGregorian(2018, 10, 1, 0, 0, 0, 0, has_year_zero=False)
 cftime.DatetimeGregorian(2018, 11, 1, 0, 0, 0, 0, has_year_zero=False)
 cftime.DatetimeGregorian(2018, 12, 1, 0, 0, 0, 0, has_year_zero=False)
 cftime.DatetimeGregorian(2019, 1, 1, 0, 0, 0, 0, has_year_zero=False)
 cftime.DatetimeGregorian(2019, 2, 1, 0, 0, 0, 0, has_year_zero=False)
 cftime.DatetimeGregorian(2019, 3, 1, 0, 0, 0, 0, has_year_zero=False)
 cftime.DatetimeGregorian(2019, 10, 1, 0, 0, 0, 0, has_year_zero=False)
 cftime.DatetimeGregorian(2019, 11, 1, 0, 0, 0, 0, has_year_zero=False)
 cftime.DatetimeGregorian(2019, 12, 1, 0, 0, 0, 0, has_year_zero=False)
 cftime.DatetimeGregorian(2020, 1, 1, 0, 0, 0, 0, has_year_zero=False)
 cftime.DatetimeGregorian(2020, 2, 1, 0, 0, 0, 0, has_year_zero=False)
 cftime.DatetimeGregorian(2020, 3, 1, 0, 0, 0, 0, has_year_zero=False)]

It would also take parameters like group_span (to include groups whose actual span is not equal to a given value) and group_contiguous (to include non-contiguous groups, or any contiguous group containing overlapping cells). A similar functionality is embedded in collapse but it works only when one has to collapse axes of the field.

Start by reading the existing subspace and collapse entry points on cf.Field, since the request describes the new cf.Field.group() method as related to both. Compare the example's seasonal and date-range selection with the requested group_span and group_contiguous parameters. Done means a documented method can select structured, non-contiguous or overlapping groups as described.