anonymous enums kill bindings

[trws]

Subject says it all, no clue why this works elsewhere but a build on toss2 died without this change.

[coveralls]

Coverage decreased (-0.03%) to 78.79% when pulling d2110d2 on trws:fix_python_bindings into 3eea25b on flux-framework:master.

[codecov-io]

Codecov Report

Merging #1354 into master will decrease coverage by 0.03%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #1354      +/-   ##
==========================================
- Coverage   78.51%   78.47%   -0.04%     
==========================================
  Files         162      162              
  Lines       29741    29741              
==========================================
- Hits        23350    23340      -10     
- Misses       6391     6401      +10

Impacted Files	Coverage Δ
src/common/libflux/ev_flux.c	97.56% <0%> (-2.44%)	⬇️
src/connectors/local/local.c	87.4% <0%> (-0.75%)	⬇️
src/common/libflux/msg_handler.c	86.64% <0%> (-0.73%)	⬇️
src/common/libflux/message.c	80.89% <0%> (-0.59%)	⬇️
src/broker/overlay.c	74.14% <0%> (-0.32%)	⬇️
src/cmd/flux-module.c	85.06% <0%> (-0.31%)	⬇️
src/common/libflux/reactor.c	93.44% <0%> (-0.29%)	⬇️
src/broker/module.c	83.79% <0%> (-0.28%)	⬇️
src/common/libflux/handle.c	83.91% <0%> (+0.24%)	⬆️
src/common/libflux/rpc.c	94.21% <0%> (+0.82%)	⬆️
... and 1 more

[garlick]

Thanks! Sorry I knew that was problematic for the python binding, but apparently it slipped by anyhow.

[trws]

No worries, just makes me wonder how it passed by all the testing. Maybe it's only an issue for certain versions or something.