Ability to cache a Transform instance

[s-boardman]

✨ Feature Request

It would be very useful to be able to cache an instance of Transform for reuse by multiple runtimes.

Motivation

My use case is that I have a load of LFric files that get generated regularly and I want to process them to the same UM like structure each time. I can't keep a python runtime open forever so would need to create a new Transform instance each time I started my processing.

Additional Context

If I wanted to use Transform for the same grid repeatedly in multiple runtimes I'd have to instantiate the same Transform instance each time which costs time. If I could then cache a Transform instance and load from that in each run time that would make life a lot easier and quicker!

[bjlittle]

@s-boardman A solid +1 on this, it's an obvious and practical feature to support.

Thanks for taking the time to raise this issue, much appreciated 👍