Make CI green - disable x86 windows and disable PPStack test on win64

[ViralBShah]

Disable conda_provided_r test - This fails on Win64, but perhaps it is ok to not test, since it is just testing threading in the R installation (not RCall itself).
Also remove x86 windows from CI (Addresses #458)

cc @DilumAluthge @randy3k. I think this may be sufficient to get CI to pass and it is ok to not test 32-bit Windows, which is anyways not a Tier 1 supported Julia platform.

[codecov]

Codecov Report

Base: 76.06% // Head: 78.01% // Increases project coverage by +1.94% 🎉

Coverage data is based on head (71c2901) compared to base (e822592).
Patch has no changes to coverable lines.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #462      +/-   ##
==========================================
+ Coverage   76.06%   78.01%   +1.94%     
==========================================
  Files          23       23              
  Lines        1571     1610      +39     
==========================================
+ Hits         1195     1256      +61     
+ Misses        376      354      -22     

Impacted Files	Coverage Δ
src/eval.jl	86.95% <0.00%> (-1.45%)	⬇️
src/render.jl	88.57% <0.00%> (ø)
src/methods.jl	67.82% <0.00%> (ø)
src/convert/default.jl	91.33% <0.00%> (+1.57%)	⬆️
src/callback.jl	81.08% <0.00%> (+2.70%)	⬆️
src/setup.jl	76.13% <0.00%> (+4.70%)	⬆️
src/macros.jl	84.37% <0.00%> (+6.25%)	⬆️
src/eventloop.jl	68.51% <0.00%> (+14.81%)	⬆️
src/io.jl	71.76% <0.00%> (+18.82%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[randy3k]

Sorry for not being able to update and maintain RCall for a while. The test is quite important here to make sure that there is no unbalanced protect / unprotect calls. These calls are used everywhere in RCall. I have used it to catch a few bugs before.

Though I am not sure why it fails on Windows.

[ViralBShah]

Thanks. Good to know. What if I just temporarily disable the test on Windows so that CI passes?

[randy3k]

I think it's fine to skip it on Windows. The mechanism is the same on unix and windows. Though I am not sure why it now fails on Windows.

[randy3k]

@DilumAluthge Do you know how the test for R_PPStackTop fails with R from conda?

[ViralBShah]

The tests have not passed ever on windows after the migration to github CI, from what I can see in the history.