[FEAT] Add better detection of Ray Job environment

[jaychia]

When running in a Ray Job, without the user invoking any Ray commands or ray.init() explicitly, the ray.is_initialized() function returns False.

This means that Daft "does not know" that it is running inside of a Ray cluster, and thus will not default to using the RayRunner. This can lead to unexpected behavior when using daft-launcher because a user must know to call daft.context.set_runner_ray().

This PR changes that behavior by attempting to look up the $RAY_JOB_ID environment variable, as a heuristic to tell whether or not it is currently running inside of a Ray job.

To test, I just ran a Ray job and called daft.context.get_context() after initializing a Daft dataframe

[codspeed-hq]

CodSpeed Performance Report

Merging #3148 will not alter performance

_{Comparing jay/ray-init-2 (32e44d0) with main (5fc9531)}

Summary

✅ 17 untouched benchmarks

[codecov]

Codecov Report

Attention: Patch coverage is 50.00000% with 2 lines in your changes missing coverage. Please review.

Project coverage is 78.99%. Comparing base (e84ed5b) to head (32e44d0).
Report is 18 commits behind head on main.

Files with missing lines	Patch %	Lines
daft/context.py	50.00%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3148      +/-   ##
==========================================
+ Coverage   78.80%   78.99%   +0.19%     
==========================================
  Files         621      634      +13     
  Lines       74809    76913    +2104     
==========================================
+ Hits        58954    60759    +1805     
- Misses      15855    16154     +299     

Files with missing lines	Coverage Δ
daft/context.py	79.75% <50.00%> (-0.25%)	⬇️

... and 75 files with indirect coverage changes

[raunakab]

Approving, but have a quick question.