[DOCS] Add overview of model providers vs functions supported

[svilupp]

Create a table in the documentation listing which model providers are supported and which functions/capabilities we support for them

[adarshpalaskar1]

Hello, can I work on this issue?
Is this the format we are looking for:

Model Provider	Supported ai* Functions
OpenAI	aigenerate
	aiembed
	aiclassify
	aiextract
	aiscan
	aiimage
Ollama	aigenerate
	aiembed
	aiscan

and so on...

Should I also include links to the examples(as done for aiembed)?

[svilupp]

That would be great!

What I imagine is a simple doc page with quick explanation of the functions (and reference link to more detailed explanation in: https://siml.earth/PromptingTools.jl/dev/how_it_works#ai*-Functions-Overview)

Then one-liner that we route ai calls through the use of subtypes of AbstractPromptSchema that explain how the data should be formatted and where to send it (eg, OpenAI models have the corresponding schema ...)

As for the format, I'm open to suggestions. I was originally thinking an overview table that includes the schema name for reference and ticks/cross/partial for the function implementation (to reduce vertical size)
abstract schema | schema | provider | aigenerate | .... |

Inspired by: https://jump.dev/JuMP.jl/stable/installation/#Supported-solvers and https://jmejia8.github.io/Metaheuristics.jl/dev/algorithms/#Algorithms

It would be good to include the catch-all implementation, eg,
AbstractOpenAISchema | CustomOpenAISchema | Any OpenAI-compatible API (eg, vLLM)* | ...
with a * footnote explaining that you need to provide URL with api_kwargs and API key.

What do you think?

Links to docstrings are an amazing idea, but perhaps let's do it as the last step. Once we nail the format of the table and information?

[adarshpalaskar1]

Thank you for outlining the structure. I think this format would be much better and easier to go through. I will start with the implementation in a PR and then work on the required changes.