Minimal support for cases with more than 2 X-points

[hrdoktorx]

The mapping data structure now stores the indices of the active and inactive X-points, and the code uses them instead of hard-coded indices 1 and 2. The active X-point is the first one to be found that is a vertex to a cell inside volumetric region 1 (the core). The inactive X-pt is chosen to be the X-point closest in Psi to the active X-point. If they do not lie on the same flux surface, the difference between DDN_TOP and DDN_BOTTOM topologies is simply whether the X-point is above or below the z=0 plane.

[jlore]

Added some comments.

Note mainly for me: Merging this will require some adaptation of my general topology PR, which I will handle after it gets reviewed by the Leuven team.

[hrdoktorx]

@jlore Sorry, I do not see your comments.

[jlore]

Ok, I tried commenting on the in-line code since I don't have this branch checked out. I'll try to @ you there to see if they are visible

[hrdoktorx]

I still see nothing.

[jlore]

I had to "submit review"

[hrdoktorx]

I would appreciate from the reviewers (@antonellozito @SVandenKerkhof @WouterDekeyser @SjoerdPR ) at least the courtesy of replying to my PRs with no more than an order of magnitude slower than I respond to theirs.
@jlore I still await your explanation for your objection to line 1067 in b2us_geo.

[jlore]

Missed that, and you are correct in your comment. My comment on 1067 can be ignored

[antonellozito]

Sorry, indeed I did not give the official OK in the PR even though I had given my informal OK in the Slack channel... apologies for that.
The changes are fine for me. As I said, these changes would be obsolete in case it is the external gridding program which already tells to B2.5 which is the topology, like I am doing right now in my forked branches which rely on GOAT, but whether/how to use this approach in the future will be widely discussed at the Code Camp. For the time being, this is fine for the official branch.

[WouterDekeyser]

Also agree with current changes - pending outcome of discussions on implementation of more generalized topologies.

[SjoerdPR]

I was asked to check if a specific case (that used to segfault before manually fixing face regions) segfaulted with this branch.

With the regular wg-release (after fixing face regions in my case), the segfault no longer happens. However, with this branch, the following error comes up (debug print statesments added right above

B2.5/src/modules/b2us_geo.F

Lines 1249 to 1250 in c550bf3

	call xertst (match_found .or. mpg%nXpt.eq.0,
	& 'No matching face found for strike point !')

).

match_found= T
nXpt= 6
match_found= F
nXpt= 6
*** XERRAB: program will stop. ***
No matching face found for strike point !
Call chain follows.
b2mndr_0
b2mn
STOP xerrab--error

I am not planning to continue with this case for my work, and it seems like a 50/50 whether this is an error with the PR or an error in my case that simply interacts poorly with the changes made in the PR. To avoid wasting time on the latter, I think the most efficient way to proceed would be to test the changes in this PR with a wide-grid case that is known to be well set-up, rather than my case.