MonthMetricsAdapter for new metrics calculators

[wpreimes]

Separate file for metric calculator adapters. Make sure that the Monthly/Seasonal metrics work for the new metric calculators, i.e. the ones from @s-scherrer

[wpreimes]

Metrics look like this, the idea is to look at e.g. seasonal variations in metrics (similar to the Rolling Metrics, but for groups)

{(('col1_name', 'col1'),
  ('col2_name', 'col2'),
  ('reference_name', 'reference')): {'gpi': array([1], dtype=int32),
  'lon': array([1.]),
  'lat': array([1.]),
  ('DJF', 'n_obs'): array([1804], dtype=int32),
  ('MAM', 'n_obs'): array([1840], dtype=int32),
  ('JJA', 'n_obs'): array([1840], dtype=int32),
  ('SON', 'n_obs'): array([1820], dtype=int32),
  ('ALL', 'n_obs'): array([7304], dtype=int32),
  ('DJF', 'snr', 'reference_name'): array([6.304975], dtype=float32),
  ('DJF', 'snr', 'col1_name'): array([6.106667], dtype=float32),
  ('DJF', 'snr', 'col2_name'): array([6.3034782], dtype=float32),
  ...

resp. for pairwise

(('reference_name', 'reference'),
  ('zcol3_name',
   'zcol3')): {'gpi': array([1], dtype=int32), 'lon': array([1.]), 'lat': array([1.]), 
('DJF',  'R'): array([0.8117465], dtype=float32), 
('DJF', 'p_R'): array([0.], dtype=float32), 
('DJF', 'BIAS'): array([-0.00691566], dtype=float32), 
('DJF', 'RMSD'): array([0.61712223], dtype=float32),
   ...