[FEATURE]Data quality in Forecasting

[kaituo]

Is your feature request related to a problem?
We have confidence in AD. Using it directly is confusing as forecasting literature has similar concept but totally different meaning.

What solution would you like?
We will use data quality instead of confidence to express the ratio of real values. The more imputation, the less data quality. We need to keep that ratio in AD and save/retrieve it in checkpoints.