Adding localized Mur\SA-Mur absorbers

[gadiLhv]

There are two additions in this pull requests:

1. Support for Mur B.C. locally (not necessary on the edge) as a primitive
   Same use cases as specified in the CSXCAD PR

• Horn antennaspattern calculation
• Coaxial to PCB transitions.
  Two types or Mur B.Cs were implemented:
• Regular first-order Mur
• Super-absorbing Mur B.C (see reference [1]). They did increase convergence rate in some cases.

2. Externally calculated modes are also integrated for V\I calculations. S-Parameters verified, as well.
   

All original functionality is maintained. All functions can be used as before, as well.

[1] Betz, Vaughn Timothy, and R. Mittra. "Absorbing boundary conditions for the finite-difference time-domain analysis of guided-wave structures." Coordinated Science Laboratory Report no. UILU-ENG-93-2243 (1993).‏

[gadiLhv]

Added two example scripts to openEMS\Python\Tutorials

[gadiLhv]

Checked that this commit\push compiles.

Ready for instructions

[thliebig]

A few initial comments and things you need to change:

• It's obvious that you took the "normal" Mur boundary condition extension as basis, make sure to fully cleanup then, e.g. comments, names, dates and such
• The new ArrayLib infrastructure has been introduced to improve array handling since you first started . Please make use of this new array approach by once again looking how it is done in the Mur-BC extension.The current approach with a '#define" is not how I usually do it (even before the ArrayLib) and running you code also creates a "double free" and subsequent segmentation fault, showing that array cleanup is not done correctly too.
• Cleanup the large commented sections
• I recommend to move the "python test code" (MSL_With_Local_Absorbers.py) into a different folder, maybe under Tests or Examples (would need to be created). Once this is all done we should create a proper Tutorial, but this is not it.
• Maybe a (rect) waveguide would be a good showcase for this ABC?

[gadiLhv]

Thank you for reviewing this so quickliy!

It's obvious that you took the "normal" Mur boundary condition extension as basis, make sure to fully cleanup then, e.g. comments, names, dates and such

Sure did! The lumped elements PR was based on Lorentz material, BTW. Will clean it up soon enough, as requested.

The new ArrayLib infrastructure has been introduced to improve array handling since you first started .

That's what I get for moving slowly :). I'll look into it.

Maybe a (rect) waveguide would be a good showcase for this ABC?

No problem, although one could claim it can be done with placing the waveguide on the boundaries... I'll figure something out. I wanted to implement a coax->waveguide transition, but the CAD work drove me nuts.

[gadiLhv]

After starting to get acquainted with the new ArrayLib, I have a question:
Unlike the original Mur BC extension, this one consists of (probably) much smaller areas to perform the operation on. What is the best approach out of these two?

1. Somehow concatenate all of the lines into one vector and input them into a single array?
2. An array of ArrayIJ, one for each boundary condition?

I don't know if this question is for @thliebig or @biergaizi, who wrote this.

Cheers

[thliebig]

I'm not sure I understand the question fully, but yes, for the 2D arrays I guess you should use the ArrayIJ instead. This will create a single/linear array, but can be efficiently accessed like a 2D array