[BUG] TypeError: ts_fit() got an unexpected keyword argument 'component_mask'

[berniwal]

Describe the bug If one creates a Timeseries with hierarchy and applies a reconciliator with a fit_transform method (here MinTReconciliator), then it always fail because it does not expect a component_mask input argument. It fails and raises the error:

TypeError: ts_fit() got an unexpected keyword argument 'component_mask'

Calling fit and transform separately works as expected.

To Reproduce

from darts.utils.timeseries_generation import linear_timeseries, constant_timeseries
from darts.dataprocessing.transformers.reconciliation import MinTReconciliator

# Create Dummy Timeseries
series = linear_timeseries(length=3)
series_2 = constant_timeseries(length=3)
stacked_series = series.stack(series_2)

# Define Hierarchy
hierarchy = {'linear': ['constant']}
stacked_series = stacked_series.with_hierarchy(hierarchy)

# Define Reconciliator
reconciler = MinTReconciliator()

# Working Setup
reconciler.fit(stacked_series)
result = reconciler.transform(stacked_series)

# Not Working Setup
result = reconciler.fit_transform(stacked_series)

Expected behavior Fit Transform works the same as calling fit and transform separately.

System:

• Python version: 3.7.12
• Darts Version: 0.20.0