Synchronizer api in V3

[jhamman]

We have been working on porting the Zarrita API into the v3 branch here. Zarrita did not include any synchronization. Options going forward are:

1. drop the synchronization api, it seems to be rarely used, especially in distributed contexts
2. port the synchronization from 2.0
3. rethink synchronization in zarr-python

(1) is the default state. (2 or 3) requires a volunteer to help make this happen.

Marking as help-wanted!

[jhamman]

@zarr-developers/python-core-devs -- it would be helpful to get some input from folks on the target outcome for v3 here.

[joshmoore]

I've been attempting to use it in anger in https://github.com/ome/ome2024-ngff-challenge/blob/main/src/ome2024_ngff_challenge/resave.py -- I agree that it feels under used/tested (#1956 (comment)). I'm in favor of a sync API though I could see a simpler one more akin to the tensorstore API: array.do_a_thing().result()