Fix memory crash/perf. issue with arima.py + exogenous regressors

[belliriccardo]

This simple patch would fix the issue described in #1006; also, by not computing the full matrices in one case and just the singular values in the other, there would also be big performance gains, depending on the time series, regressors, etc.

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
0 out of 2 committers have signed the CLA.

:x: belliriccardo
:x: elephaint
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[elephaint]

@belliriccardo Thanks for your contribution!

[belliriccardo]

Hey @elephaint ! Sorry I'm late, I've just seen this. Thanks for the feedback; I didn't know about nbdev_export, but I can see that you already merged the changes. Do I still need to run the command or do something else?