Avoid materializing intermediate adjacency list

[simsurace]

For graphs where the adjacency is not stored, this can lead to a lot of allocations when iterating over edges.

Before PR:

julia> g
{58089, 120148} directed simple Int32 graph

julia> typeof(g)
StaticGraphs.StaticDiGraph{Int32, Int32}

julia> @time collect(Graphs.edges(g));
 21.959129 seconds (330.20 k allocations: 260.015 GiB, 27.19% gc time, 0.21% compilation time)

After PR:

julia> @time collect(Graphs.edges(g));
  0.000396 seconds (3 allocations: 938.844 KiB)

[KristofferC]

Couldn't this be done in the iterate above as well?

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 97.27%. Comparing base (48c42c7) to head (4efde07).

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #372      +/-   ##
==========================================
- Coverage   97.28%   97.27%   -0.01%     
==========================================
  Files         118      118              
  Lines        6876     6874       -2     
==========================================
- Hits         6689     6687       -2     
  Misses        187      187              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[gdalle]

Gotta love a 12-line PR with a times-one-gazillion speedup