Add Arize Phoenix tracing instructions to the observability docs page

[Jgilhuly]

Hi DSPy team - I was working through a tutorial on prompt optimization using DSPy for Arize Phoenix, and noticed the docs page on observability just showed MLFlow. I, somewhat selfishly, figured it could be good to have other observability options on there, similar to what other coming tools do (e.g. litellm).

Phoenix also will trace the optimization process when compiling a teleprompter, which I thought could be helpful on top of what MLFlow shows.

Made this PR to include Arize Phoenix as an option, and figured others could follow suit.

Hope you don't mind! Lmk of any desired changes or questions

[okhat]

Hey @Jgilhuly ! Sure thing but is the Phoenix integration up to date? We want to be conservative in recommending up to date integrations. Many of the older ones from prior to 2.4 may not work anymore out of the box.

[mikeldking]

Hey @Jgilhuly ! Sure thing but is the Phoenix integration up to date? We want to be conservative in recommending up to date integrations. Many of the older ones from prior to 2.4 may not work anymore out of the box.

Hey @okhat - There was a recent fix for 2.6 and above (Arize-ai/openinference#1249). Is that what you were looking for?

[Jgilhuly]

If it helps - the gif there was pulled from the test app I'm using, with:
DSPy Version: 2.6.3
openinference-instrumentation-dspy Version: Version: 0.1.17

So seems to be all working there

[Jgilhuly]

Hey @okhat - let us know if there's any changes you'd like here!