>>> spec = ts.Spec({
...     'driver': 'zarr',
...     'kvstore': 'memory://',
... })
>>> spec.update(shape=[100, 200], dtype=ts.uint32)
>>> cast_spec = ts.cast(spec, ts.float32)
>>> cast_spec
Spec({
  'base': {
    'driver': 'zarr',
    'dtype': 'uint32',
    'kvstore': {'driver': 'memory'},
    'schema': {
      'domain': {'exclusive_max': [100, 200], 'inclusive_min': [0, 0]},
    },
  },
  'driver': 'cast',
  'dtype': 'float32',
  'transform': {
    'input_exclusive_max': [[100], [200]],
    'input_inclusive_min': [0, 0],
  },
})
>>> cast_spec[30:40, 20:25].base
Spec({
  'driver': 'zarr',
  'dtype': 'uint32',
  'kvstore': {'driver': 'memory'},
  'schema': {'domain': {'exclusive_max': [100, 200], 'inclusive_min': [0, 0]}},
  'transform': {
    'input_exclusive_max': [40, 25],
    'input_inclusive_min': [30, 20],
  },
})
>>> downsampled_spec = ts.downsample(spec,
...                                  downsample_factors=[2, 4],
...                                  method='mean')
>>> downsampled_spec
Spec({
  'base': {
    'driver': 'zarr',
    'kvstore': {'driver': 'memory'},
    'schema': {
      'domain': {'exclusive_max': [100, 200], 'inclusive_min': [0, 0]},
    },
    'transform': {
      'input_exclusive_max': [[100], [200]],
      'input_inclusive_min': [0, 0],
    },
  },
  'downsample_factors': [2, 4],
  'downsample_method': 'mean',
  'driver': 'downsample',
  'dtype': 'uint32',
  'transform': {
    'input_exclusive_max': [[50], [50]],
    'input_inclusive_min': [0, 0],
  },
})
>>> downsampled_spec[30:40, 20:25].base