feat: Moonshot supported function call & function call unit tests

[mxsl-gr]

PR content:

1. Moonshot supported function call
2. function call unit tests

related issue: #1058

[tzolov]

HI @mxsl-gr and thank you for the update.
As we are deprecating AbstractFunctionCallSupport in favor of AbstractToolCallSupport would you be interested to re-work your PR to support AbstractToolCallSupport?

For examples, you can check the OpenAI/AzureOpenAI/Anthropic/Mistral/Ollama/VertexAIGemini function calling implementations, or ask me for support ;)
Would appreciate your help as you own those models.

[mxsl-gr]

hi, @tzolov
it worked for now, and this PR has been squash and force pushed

[markpollack]

@mxsl-gr thanks, sorry I haven't gotten to merging this yet

[mxsl-gr]

@mxsl-gr thanks, sorry I haven't gotten to merging this yet

hi @markpollack , is ok, i went on a trip and just back now. please let me know if there's anything else to adjust.

other some PRs related to function calls: #1116, #1163

[markpollack]

This test fails for me. It looks like it wants to make the calls sequentially or something

java.lang.AssertionError: 
Expecting actual:
  "The current weather in San Francisco is 30.0°C. Now, I will check the weather in Tokyo."
to contain:
  ["30", "10", "15"]
but could not find:
  ["10", "15"]
 

[markpollack]

I'm merging this despite test failure as this PR has improved the support so significantly. We can circle back to figure out what is going on with the function calling failure in the test. The streaming function calling test did pass in my testing.

[markpollack]

merged in 611c949