Cube arithmetic points/bounds mismatch errors

[rcomer]

Following on from the outstanding cube arithmetic work, some thoughts/ideas about the exception that gets raised if our coordinate points or bounds don't match.

1. [ ] 1. As already noted in #3774, it would be good to check all the coordinates and then raise an exception indicating the mismatches in all of them. This would allow the user to fix the issue all at once, rather than potentially going through several run code --> fix coords loops.

2. [ ] 2. It would be handy to get an indication of how different the coordinates are. E.g., for example 3 in my comment here, if the error message tells my that all the points are out by three hours, I can think "that's fine given what I know about the data" and do some wrangling to get the points checking to pass. If, on the other hand, I get a message saying that latitude points are out by "up to 180 degrees", I know something has gone properly wrong. This happened to me previously as I didn't realise that Windspharm enforces a latitude direction.

I'd be happy to have a go at these myself, assuming I can find the time.

[bjlittle]

@rcomer Awesome, thanks 👍

With regards to each of the points you raised:

1. I think that this is completely possible to do. I totally agree on the benefit of getting a single aggregated exception, rather than having to iterate through possibly several expensive runs, each of which raise an exception.
2. This might need some further discussion to find a practical approach, but we can cross that bridge when we come to it

[rcomer]

Just to note that, back in April, I did start work on a branch for [1] but got a bit stuck. Happy to resurrect it with some help.