Constraint selection can be unreasonably slow for time coords

[pp-mo]

Discovered that doing a date selection for a given year + month based on categorised coordinates, with an operation like month_data = cube.extract(Constraint(year=y, month=m)).

It seems much quicker to do

indices = (cube.coord('year').points == y) & (cube.coord('month').points == m)
month_data = cube[indices]

In short, this is because operation is not vectorised (unlike the above index-calculating solution).
This is worse for time coordinates, as a unit.num2date() is applied individually to each time cell (points + bounds), before calling the discriminator function.

This could definitely be fixed, within the existing API, at least for the non-user-function coordinate selection styles ("equals-point-value" and "in-list-of-values").
A useful addition would be an extension to the extract API to allow user functions to also be called with vector info.
E.G. either a recognised property visible on the function, or a keyword like vectorise_functions=True ?

TODO:
add a simple testcase demonstrating the performance hit

[github-actions]

In order to maintain a backlog of relevant issues, we automatically label them as stale after 500 days of inactivity.

If this issue is still important to you, then please comment on this issue and the stale label will be removed.

Otherwise this issue will be automatically closed in 28 days time.

[github-actions]

This stale issue has been automatically closed due to a lack of community activity.

If you still care about this issue, then please either:

• Re-open this issue, if you have sufficient permissions, or
• Add a comment pinging @SciTools/iris-devs who will re-open on your behalf.

[rcomer]

This was done at #4969 🎉