Case aggregation fails for partially overlapping data

### Describe the bug

The all case aggregation seems to fail for cases where some of the data overlap for validity times and others don't.
We think this is a bug in collapse.collapse_by_validity_time.
For now the advice is to switch off the time_series_surface_field_aggregation for all cases as it assumes no overlapping time intervals in the rose conf file.

### How to reproduce

CSET version 25.7.0 (2025-07-17)
CSET_CASE_DATES=["20220719T0000Z", "20220817T0000Z", "20220913T0000Z", "20221012T0000Z", "20221102T0000Z", "20221103T0000Z", "20221107T0000Z", "20221111T0000Z", "20221129T0000Z", "20221209T0000Z", "20221211T0000Z", "20221215T0000Z", "20221216T0000Z", "20221218T0000Z", "20230105T0000Z", "20230117T0000Z", "20230118T0000Z", "20230121T0000Z", "20230124T0000Z", "20230127T0000Z", "20230202T0000Z", "20230527T0000Z", "20230607T0000Z", "20230610T0000Z", "20230618T0000Z", "20230621T0000Z", "20230622T0000Z", "20230628T0000Z", "20230629T0000Z", "20230703T0000Z", "20230708T0000Z", "20230709T0000Z", "20230715T0000Z", "20230926T0000Z", "20231018T0000Z", "20231101T0000Z", "20231112T0000Z", "20231209T0000Z", "20231226T0000Z", "20240102T0000Z", "20240121T0000Z", "20230405T0000Z", "20220616T0000Z", "20220330T0000Z", "20210213T0000Z", "20191014T0000Z", "20190725T0000Z", "20221011T0000Z", "20220616T0000Z"]

lfric: %Y%m%dT%H%MZ_UK_672x672_1km_RAL3P3_lfric_slam_[dpt]*
um: %Y%m%dT%H%MZ_UK_672x672_1km_RAL3P3_pdiag[abcd]*
CSET_UK_672x672_1km_RAL3P3_708p1

Steps to reproduce the behaviour:

1.

2.
3.

### Expected behaviour

CSET/src/CSET/operators/plot.py", line 1396, in plot_line_series
raise ValueError("Cube must be 1D.")
ValueError: Cube must be 1D.
ERROR conda.cli.main_run:execute(125): `conda run run_cset_recipe.py` failed. (See above for error)