Add ni and nj as dimensions to be corrected

[markusritschel]

As a remark:
Sometimes, lon and lat are 2D coordinates depending on ni and nj, respectively. For example,

<xarray.Dataset>
Dimensions:    (bnds: 2, ni: 320, nj: 384, nvertices: 4, time: 1980)
Coordinates:
  * time       (time) object 1850-01-15 12:00:00 ... 2014-12-15 12:00:00
    lon        (nj, ni) float64 ...
    lat        (nj, ni) float64 ...
  * ni         (ni) int32 1 2 3 4 5 6 7 8 9 ... 313 314 315 316 317 318 319 320
  * nj         (nj) int32 1 2 3 4 5 6 7 8 9 ... 377 378 379 380 381 382 383 384
. . .

I added the entries for ni and nj in the rename_dict pretty much in the front of the list since, otherwise, it could lead to unwanted behavior.
The current routine checks only for the first appearance of the key in that list. If ni was placed at the end of the list, the routine would stop at lon (setting the trigger=True and would not even check for ni. In the above example, this would rename the 2D coordinate lon (nj, ni) to x (ni, nj).
The desired behavior, however, is:

    lon        (y, x) float64 ...
    lat        (y, x) float64 ...
  * x          (x) int32 1 2 3 4 5 6 7 8 9 ... 313 314 315 316 317 318 319 320
  * y          (y) int32 1 2 3 4 5 6 7 8 9 ... 377 378 379 380 381 382 383 384

So, one must really keep the order in mind.
Maybe this should be considered when revising the routine at some point.

[codecov]

Codecov Report

Merging #51 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master      #51   +/-   ##
=======================================
  Coverage   59.92%   59.92%           
=======================================
  Files           6        6           
  Lines         564      564           
=======================================
  Hits          338      338           
  Misses        226      226           

Impacted Files	Coverage Δ
cmip6_preprocessing/preprocessing.py	78.94% <ø> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update d1c5f2f...e0892b4. Read the comment docs.

[jbusecke]

Ah yes. That is a good point. I was aware of that and aranged these names last, but I agree this is brittle. I will open another issue to track this and think about a way to implement this in a better way.

[jbusecke]

This looks good to me. Dont forget to add an entry in doc/whats-new.rst and then we can get this merged.

Thanks again.

[markusritschel]

This looks good to me. Dont forget to add an entry in doc/whats-new.rst and then we can get this merged.

Thanks again.

Sure. Where in the doc file shall I place it? Under v0.2.0 (unreleased) > Internal changes?

[markusritschel]

Hope this is on the right place?

[jbusecke]

Yes. Awesome. Will merge when the tests have all passed.

[jbusecke]

Travis is acting weird. Ill close this and reopen to trigger a new build

[jbusecke]

This is quite frustrating. I will wait a bit, to see if there are some outages at either gh or travis that might clear up...
Sorry for the delay.

[markusritschel]

This is quite frustrating. I will wait a bit, to see if there are some outages at either gh or travis that might clear up...
Sorry for the delay.

Have you tried doing the CI with Github actions?

[jbusecke]

Not yet, but this experience here certainly is a compelling reason to switch...unbelievable. Ill force the merge, since these are very small changes and the PR test went trough.