Use julia_gap and gap_julia in more places

[fingolfin]

... for consistency in the auto-conversion behaviour of JuliaInterface versus
LibGAP.jl, and for ease of use.

There is more work to be done here, though, and we need to properly
document the rules for when and where which conversion is applied,
and why.

[codecov]

Codecov Report

Merging #127 into master will increase coverage by 0.01%.
The diff coverage is 94.11%.

@@            Coverage Diff             @@
##           master     #127      +/-   ##
==========================================
+ Coverage   54.33%   54.34%   +0.01%     
==========================================
  Files          44       44              
  Lines        2632     2622      -10     
==========================================
- Hits         1430     1425       -5     
+ Misses       1202     1197       -5

Impacted Files	Coverage Δ
JuliaInterface/gap/JuliaInterface.gi	92.59% <100%> (+0.13%)	⬆️
JuliaInterface/src/calls.c	94.26% <100%> (+0.27%)	⬆️
JuliaInterface/gap/JuliaInterface.gd	100% <100%> (ø)	⬆️
JuliaInterface/src/JuliaInterface.c	88.34% <83.33%> (+0.24%)	⬆️
JuliaInterface/read.g	92.3% <0%> (ø)	⬆️
JuliaInterface/init.g	72.09% <0%> (ø)	⬆️
JuliaInterface/src/convert.c	95% <0%> (+15%)	⬆️

[sebasguts]

This looks good.

Note that this changes some things, and might cause some problems with JuliaExperimental.

How do we want to proceed: Change a lot in JuliaInterface and then make JuliaExperimental catch up later?

@ThomasBreuer what do you think?

[fingolfin]

My personal stance on this is this: This PR updates the existing tests for JuliaExperimental, so code covered by that is fine. All the other code in JuliaExperimental has a high chance of already being completely broken anyway. IMHO the only way to effectively move on with this project is that for all code, we either add comprehensive tests (then whoever makes changes has a chance of discovering anything they break, and subsequently can fix it), or else we remove that code eventually.

[sebasguts]

Sorry, but some tests are failing. Should I correct them?

[ThomasBreuer]

@sebasguts Yes, merging proposed changes in JuliaInterface is fine, independent of what they do with JuliaExperimental.
@fingolfin Adjusting tests in JuliaExperimental to changes in JuliaInterface is perhaps not recommended in general; the code is experimental, and it may well be that changes in JuliaInterface make bigger changes in JuliaExperimental sensible.

[fingolfin]

Those test failures were real bugs, yay to the new tests that helped uncover them before merging!

@ThomasBreuer of course I am assuming that any adjustment to JuliaExperimental tests would involve some reasonable thinking, and questioning of what is there. Still, for tests to be most effective, it is important that they all pass (resp. that tests which are expected to fail are marked as such).