[Core] Fix update_usage_summary when response doesn't have usage attribute

[yiranwu0]

Why are these changes needed?

Related issue number

Close #1002 #984

Checks

• I've included any doc changes needed for https://microsoft.github.io/autogen/. See https://microsoft.github.io/autogen/docs/Contribute#documentation to build and test documentation locally.
• I've added tests (if relevant) corresponding to the changes introduced in this PR.
• I've made sure all auto checks have passed.

[codecov-commenter]

Codecov Report

Attention: 4 lines in your changes are missing coverage. Please review.

Comparison is base (1508f53) 30.82% compared to head (27a04b2) 64.43%.

Files	Patch %	Lines
autogen/oai/client.py	60.00%	4 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #1008       +/-   ##
===========================================
+ Coverage   30.82%   64.43%   +33.61%     
===========================================
  Files          30       30               
  Lines        4033     4043       +10     
  Branches      915      964       +49     
===========================================
+ Hits         1243     2605     +1362     
+ Misses       2711     1167     -1544     
- Partials       79      271      +192     

Flag	Coverage Δ
unittests	64.33% <60.00%> (+33.56%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[qingyun-wu]

@kevin666aa The PR LGTM overeall. Have you tested if this sufficiently address the issue raised in #1002 and #984. Can they be closed with this PR? Is it possible to add a test here?

[yiranwu0]

@kevin666aa The PR LGTM overeall. Have you tested if this sufficiently address the issue raised in #1002 and #984. Can they be closed with this PR? Is it possible to add a test here?

I cannot replicate the error, so I can't put down a test for this. The bug is happening because response.usage may be a nonetype due to the use of a local model. The code I added in the try catch block would make sure that response.usage can be referenced, and it is not none type. Further checking of response.prompt_tokens is also added. So I think it would be sure to resolve the two issues.

I can ask people to test it out to see if it works.

[sonichi]

openai 1.5 support has been added by #1044 + #1071
There is no need to limit the openai version in this PR.

[ekzhu]

We can merge this once the conflict has been resolved. @kevin666aa

[ekzhu]

@kevin666aa failing tests.

[yiranwu0]

@kevin666aa failing tests.

@kevin666aa failing tests.

Hello @ekzhu, can you help me run the openai test again?

[sonichi]

#1110 needs to be merged first to prevent test failures.