Set deadline for making iris.experimental.ugrid non-experimental

[trexfeathers]

No description provided.

[trexfeathers]

See #5725

[trexfeathers]

Existing requests that are (or might be) breaking changes

• Better API for accessing coordinates on a Mesh #5993
  Ideally this would only add functionality rather than changing anything existing, but it's still being specced out.
• Support global enable of ugrid loading #5940
  We have defended our design decision based on still being experimental when the change is made.
• Refactor Mesh to MeshXY #6052
  Specifically designed as a quick way to get the breaking change in quickly while still experimental, allowing the larger proportion of the work (Should `Mesh` class allow for any number of node coordinates? #5994) to take place at any time.

Other requests I have looked over could deliberately be implemented in a way that would avoid breaking most existing user scripts.

[trexfeathers]

Future feedback that might require breaking changes

I can obviously only speculate. Every month Iris' mesh API gets exercised more, but this is by definition early adopters. We will only know if it is truly fit for purpose once it is used operationally, and that won't happen while it is marked as experimental (indeed we have already had conversations with someone about this being a problem).

We do know from the volume of issues and offline conversations that many people have already used Iris' mesh API and it has worked for them.

Any date we set on this basis will be arbitrary. There is no concrete argument for delay here - there isn't a key person we're waiting to hear from or anything like that.

[trexfeathers]

3.10 (now) is the best time

We're in a developer sprint with multiple experts available to tackle this, and coordination of potential conflicts.