Switch of collecting data in non-parallel run.

[blattms]

Might save some time, space, and suprises.

Completes #137. @andlaus merges faster than I commit.

[blattms]

and gets rid of the assertion mentioned in #137

[andlaus]

push green if you deem it to be ready.

[blattms]

I will not self merge. Otherwise I cannot complain about that in the future anymore.

[atgeirr]

Now that is a deadlock...

[andlaus]

I will not self merge. Otherwise I cannot complain about that in the future anymore.

ok, I'll merge, but I refuse to be responsible for any breakage.

[atgeirr]

ok, I'll merge, but I refuse to be responsible for any breakage.

I think a maintainer will always at least share the responsibility for handling a situation where things break, unless bullied by the PR author to merge :-) (I do not think that is the case here.)

More to the point: did you check this @andlaus, by reading or running the code, or both?

[andlaus]

Well, I think that the author of a patch has a certain responsibility to test and self-review the stuff before opening a PR. (at least myself has been doing this for my PRs during the last few years as you've doubtlessly noticed. ;))

I looked at the code, and nothing seemed to be obviously wrong, but I usually only compile stuff in optimized mode and I do not run flow_ebos in parallel because the latter does not work yet anyway. (the first point was also true for #137, BTW.)