add Pluto tutorials

[CarloLucibello]

A step toward fixing #168.

Also adding a few utilities

• Graphs.Graph(::GNNGraph) conversion
• has_isolated_nodes(g)
• to_unidirected(g) which removes the reves edges and can be useful for plotting an undirected graph which is bidirected as an undirected one.

[codecov]

Codecov Report

Merging #172 (a95178a) into master (c7d0cbb) will increase coverage by 0.68%.
The diff coverage is 92.00%.

@@            Coverage Diff             @@
##           master     #172      +/-   ##
==========================================
+ Coverage   85.33%   86.01%   +0.68%     
==========================================
  Files          15       15              
  Lines        1316     1323       +7     
==========================================
+ Hits         1123     1138      +15     
+ Misses        193      185       -8     

Impacted Files	Coverage Δ
src/GNNGraphs/utils.jl	91.02% <66.66%> (+6.30%)	⬆️
src/GNNGraphs/query.jl	93.64% <100.00%> (+0.03%)	⬆️
src/GNNGraphs/transform.jl	96.91% <100.00%> (+0.17%)	⬆️
src/GNNGraphs/convert.jl	90.08% <0.00%> (-0.32%)	⬇️
src/GNNGraphs/sampling.jl	100.00% <0.00%> (ø)
src/layers/conv.jl	77.31% <0.00%> (+0.07%)	⬆️
src/GNNGraphs/gnngraph.jl	75.00% <0.00%> (+2.50%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update ec6951f...a95178a. Read the comment docs.

[CarloLucibello]

There are some problems that I don't understand in the github action compiling the pluto notebooks. I'll leave them out for the time being and solve in a followup PR

[rikhuijzer]

There are some problems that I don't understand in the github action compiling the pluto notebooks. I'll leave them out for the time being and solve in a followup PR

Cool to see that you're going to use PlutoStaticHTML. Let's get it working

You have set

use_distributed=false

From the PlutoStaticHTML docs:

Beware that use_distributed=false will not work with Pluto's built-in package manager.

So, you can fix it in two ways. The first option is to explicitly install the packages manually. So, something like:

begin
    using Pkg
    Pkg.activate(; temp=true)
    packages = [
        PackageSpec(; path=dirname(dirname(@__DIR__))),
        PackageSpec(; name="Flux", version="0.13")
        [...]
    ]
    Pkg.add(packages)
end

begin
    using GraphNeuralNetworks
    using Flux
    ...
end

I would personally do this because it is very explicit and therefore easy to maintain.

An alternative would be to set use_distributed=true and use Pluto's built-in package manager. However, then you need to figure out how to install the latest GraphNeuralNetworks in your notebook. This is particularly difficult when you are on a PR. Or do you know a way to install a relative path via Pluto's package manager? Maybe that's possible. I haven't checked in detail