Rewrite "why cubed?" for users

[TomNicholas]

The current text on why cubed? is written at a very high level. It's very interesting for people who think about different distributed systems, but it's not super clear to new users who just want to know what cubed can do for them.

Perhaps we should move the existing text to e.g. design/motivation, and rewrite this intro text to instead emphasise:

• cubed is a drop-in replacement for dask.array
• it tells you if your computation would run out of memory
• it is designed to be very robust to failures, and be resumable
• it doesn't require you to make many (if any) up-front choices about configuration / cluster size
• (Optional) IF you want to run in the cloud it is very horizontally scalable and stateless / serverless

[tomwhite]

Thanks for opening this @TomNicholas!

Perhaps we should move the existing text to e.g. design/motivation

Definitely. It's mainly of historical interest, so should be moved down to the bottom of the nav links (I noticed that DuckDB does this).

The intro text at https://cubed-dev.github.io/cubed/ (and https://github.com/cubed-dev/cubed/blob/main/README.md) needs updating, as you say.

• "Zarr for storage" is a bit vague. It's about how Zarr is used for intermediate storage, which is an implementation detail that is probably not a top-level bullet point. Being able to read and write the user's Zarr data is more relevant.
• "Multiple serverless runtimes: Python (in-process)" - change to add out-of-core/processes.
  • Maybe say something like "Multiple runtimes: local (...), serverless (...), cluster (Beam, Dask)"
• Add something about accelerators (JAX)

These are just a few ideas. I think it would be really useful to collaborate on re-writing the intro.