Normalising time coordinate information when loading LFRic and UM

[jwarner8]

LFRic time dimension data looks very different to the UM and currently the collapse by hour, lead time and validity time operators in conjunction with aggregation across case studies currently doesn't work using LFRic data as it does not contain forecast_reference_time or forecast_period.

For example in UM;

DimCoord :  time / (hours since 1970-01-01 00:00:00, standard calendar)
    points: [
        2022-07-13 03:00:00, 2022-07-13 06:00:00, ...,
        2022-07-14 21:00:00, 2022-07-15 00:00:00]
    shape: (16,)
    dtype: float64
    standard_name: 'time'

For example in LFRic;

AuxCoord :  time / (seconds since 2022-07-13 00:00:00, standard calendar)
    points: [
        2022-07-13 00:30:00, 2022-07-13 01:30:00, 2022-07-13 02:30:00,
        2022-07-13 03:30:00, 2022-07-13 04:30:00, 2022-07-13 05:30:00]
    bounds: [
        [2022-07-13 00:00:00, 2022-07-13 01:00:00],
        [2022-07-13 01:00:00, 2022-07-13 02:00:00],
        [2022-07-13 02:00:00, 2022-07-13 03:00:00],
        [2022-07-13 03:00:00, 2022-07-13 04:00:00],
        [2022-07-13 04:00:00, 2022-07-13 05:00:00],
        [2022-07-13 05:00:00, 2022-07-13 06:00:00]]
    shape: (6,)  bounds(6, 2)
    dtype: float64
    standard_name: 'time'
    long_name: 'Time axis'
    var_name: 'time_centered'
    attributes:
        time_origin  '2022-07-13 00:00:00'