Using a more robust "reflection_with_llm" summary method

[qingyun-wu]

Why are these changes needed?

1. The current way of getting a summary under the reflection_with_llm option is not robust (not working well with function/tool call). This change re-uses code from ConversableAgent's generate_oai_reply, which should be more robust.

2. Default behavior: if "summary_method" is not specified (the default setting), uses last msg as the summary.

3. Enhancing tests to cover result return and summary

Related issue number

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 (e0fa6ee) 35.19% compared to head (51bcf23) 68.70%.

Files	Patch %	Lines
autogen/agentchat/conversable_agent.py	80.00%	3 Missing and 1 partial ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #1575       +/-   ##
===========================================
+ Coverage   35.19%   68.70%   +33.51%     
===========================================
  Files          44       44               
  Lines        5328     5323        -5     
  Branches     1236     1303       +67     
===========================================
+ Hits         1875     3657     +1782     
+ Misses       3299     1317     -1982     
- Partials      154      349      +195     

Flag	Coverage Δ
unittests	68.66% <80.00%> (+33.47%)	⬆️

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.

[ekzhu]

Consider improving the formatting of ChatResult class's __str__ and __repr__, right now it is hard to read.