Add support to masking only specified variables during training

[IncubatorShokuhou]

1. Feature description

Add support to masking only specified variables during training, rather than randomly masking all variables by default.

2. Motivation

In PyPOTS, during the training process, variables are randomly masked by default.
But in real-world time series data, many covariates do not have missing values and therefore do not require masking or prediction. For example, in the field of meteorology, if we use variables from numerical weather models as covariates, these variables cannot have missing values because numerical weather forecasts are guaranteed to be available every day. In this case, this default behavior can be unnecessary and inefficient.
So, a feature is requested to be added that allows users to specify which variables should be masked during training. This would enable more targeted and efficient training processes, especially for datasets where only certain variables have missing values that need to be handled.

3. Your contribution

[github-actions]

Hi there 👋,

Thank you so much for your attention to PyPOTS! You can follow me on GitHub
to receive the latest news of PyPOTS. If you find PyPOTS helpful to your work, please star⭐️ this repository.
Your star is your recognition, which can help more people notice PyPOTS and grow PyPOTS community.
It matters and is definitely a kind of contribution to the community.

I have received your message and will respond ASAP. Thank you for your patience! 😃

Best,
Wenjie