fix: improve retry behavior with resources at MG scope

[matt-FFFFFF]

As per internal thread:

• Improved 403 handling for MG resources to include child resources
• Fix bug in get after put retry timeout so the environment variable works properly
• Added debug output to retry

I could also pass the regex to the get after put retry, but since the body is nil then the retry func catches the issue where we get a 403. WDYT @ms-henglu

[ms-henglu]

Hi @matt-FFFFFF ,

Thank you for the PR, LGTM.

I could also pass the regex to the get after put retry, but since the body is nil then the retry func catches the issue where we get a 403. WDYT

I think it's better to allow the user specified retry block to control retry policy of the GET after PUT. Maybe combine both 404 or empty body retry with the user configured retry policy? By doing that, user could also specify the max elapsed time in the configuration and get rid of the environment variables.

[matt-FFFFFF]

Hi @ms-henglu I have rebased and updated this PR to pass the regex messages to the getafterput client.

The max time was omitted from the retry object because we rely on the resource timeout value (context deadline). I think having two timeouts could be confusing?

WDYT?

[ms-henglu]

Thanks for the PR, LGTM.