Should we speed-up xcdat import?

[pochedls]

It feels like xcdat is slow on import (i.e., import xcdat), though I recognize it only takes ~6s. This is much slower than other libraries (xarray, pandas, and scipy clock in under 1s). They are probably loading sub-modules or dependencies on demand. Some questions:

• Do others also feel xcdat is a little too sluggish on import?
• Maybe we should profile this to see what is causing the slowdown? Are we importing the lowest level of modules needed?
• Should we defer imports of dependencies until we need them? Or use lazy_loader?

Thoughts?

[shoyer]

This is probably a better issue for the xCDAT repo than Xarray -- feel free to open another issue is there's a legitimate Xarray issue, though!
https://github.com/xCDAT/xcdat

[pochedls]

Sorry! I mistakenly posted this to the wrong repository.