Resolve ML cluster failures

[charlesbluca]

It looks like something changed in upstream Dask such that we're now seeing issues when trying to use the local ML wrapper functions introduced in #832 and #855 in an independent cluster; this PR wraps these relevant function usages with make_pickable_without_dask_sql as is done with other functions passed directly off to Dask, and adds the skip_if_external_scheduler marker to relevant tests that would otherwise need scikit-learn.

cc @sarahyurick

[codecov-commenter]

Codecov Report

Merging #957 (8b75dd7) into main (f3fbdd0) will increase coverage by 0.14%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #957      +/-   ##
==========================================
+ Coverage   76.62%   76.76%   +0.14%     
==========================================
  Files          73       73              
  Lines        4115     4119       +4     
  Branches      751      751              
==========================================
+ Hits         3153     3162       +9     
+ Misses        792      783       -9     
- Partials      170      174       +4     

Impacted Files	Coverage Δ
dask_sql/physical/rel/custom/wrappers.py	64.59% <100.00%> (+0.52%)	⬆️
dask_sql/_version.py	32.76% <0.00%> (+1.41%)	⬆️

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[sarahyurick]

Interesting, I wasn't familiar with make_pickable_without_dask_sql but makes sense to me! Thanks

[ayushdg]

changes lgtm! Thanks