add "address_remap" feature to RedisCluster

[kristjanvalur]

Pull Request check-list

• Does $ tox pass with this change (including linting)?
• Do the CI tests pass with this change (enable it first in your forked repo and wait for the github action build to finish)?
• Is the new or changed code fully tested?
• Is a documentation update included (if this change modifies existing APIs, or introduces new ones)?
• Is there an example added to the examples folder (if applicable)?
• Was the change added to CHANGES file?

Description of change

A new parameter, address_remap is added to both both RedisCluster and asyncio.RedisCluster. This allows remapping the addresses that the cluster nodes advertise about themselves, to something which is accessible to a client. For example, the nodes might sit behind a NAT, or a reverse proxy, that the nodes themselves don't know about. The address_remap will ensure that the RedisCluster object will connect to the remapped addresses when connecting to the cluster.

This PR could be considered a precursor to #2695 and #2706 since both of these provide this functionality in order to facilitate unit-tests, albeit only for the asyncio.RedisCluster. Accepting this PR would allow those other PRs to be simplified.

Reference

This feature is inspired by the host_port_remap feature described here: https://redis-py-cluster.readthedocs.io/en/stable/client.html
but we use a more flexible callback based mechanism rather than a dictionary of rules.

[codecov-commenter]

Codecov Report

Patch coverage: 98.20% and project coverage change: -0.02 ⚠️

Comparison is base (ac15d52) 92.24% compared to head (0e7447a) 92.23%.

📣 This organization is not using Codecov’s GitHub App Integration. We recommend you install it so Codecov can continue to function properly for your repositories. Learn more

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2726      +/-   ##
==========================================
- Coverage   92.24%   92.23%   -0.02%     
==========================================
  Files         116      116              
  Lines       29928    30093     +165     
==========================================
+ Hits        27606    27755     +149     
- Misses       2322     2338      +16     

Impacted Files	Coverage Δ
tests/test_asyncio/test_cluster.py	97.15% <97.01%> (-0.33%)	⬇️
tests/test_cluster.py	96.65% <98.82%> (-0.01%)	⬇️
redis/asyncio/cluster.py	91.91% <100.00%> (+0.09%)	⬆️
redis/cluster.py	90.49% <100.00%> (+0.07%)	⬆️

... and 5 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[chayim]

Hey @kristjanvalur Just wanted to let you know that the failing test appears to be a red herring. Not sure if you're ready for review and whatnot, but I did want to let you know ;)

[kristjanvalur]

Hey @kristjanvalur Just wanted to let you know that the failing test appears to be a red herring. Not sure if you're ready for review and whatnot, but I did want to let you know ;)

Thanks, yes, I had figured as much. Timeout-based tests are notoriously (in my experience) prone to failure in this repo, unfortunately.

Anyway: I think this is ready for review. I was thinking that it would probably be best to get this in (or not) before merging my other prs, since fixing the test_cwe_404.py for cluster mode relies on this kind of functionality. I've updated the description to that effect.