Disttributed MLForecast needs a pandas data frame for predict when using exog features

[janvenu-tw]

What happened + What you expected to happen

The DistributedMLForecast feature is to allow the development of scalable models, we have to convert the exog features to pandas before the predict functionality, then its sure to run out of memory. Please help avoid the conversion to pandas in the predict

Versions / Dependencies

mlforecast version 1.02

Reproduction script

This exact code does not work if the soze pf the data goes up The example code on MLforecast page

Issue Severity

None