Fix crash if the minimum and maximum depth are the same.

[MFraters]

use the top temperature in that case.

[github-actions]

Benchmark	Main	Feature	Difference (99.9% CI)
Slab interpolation simple none	1.020 ± 0.003 (s=451)	1.020 ± 0.004 (s=434)	-0.1% .. +0.1%
Slab interpolation curved simple none	1.017 ± 0.004 (s=436)	1.024 ± 0.004 (s=448)	+0.7% .. +0.8%
Spherical slab interpolation simple none	1.172 ± 0.008 (s=381)	1.173 ± 0.008 (s=389)	-0.1% .. +0.2%
Slab interpolation simple curved CMS	1.059 ± 0.004 (s=426)	1.060 ± 0.003 (s=426)	-0.1% .. +0.1%
Spherical slab interpolation simple CMS	1.551 ± 0.009 (s=299)	1.550 ± 0.009 (s=284)	-0.2% .. +0.1%
Spherical fault interpolation simple none	1.173 ± 0.011 (s=367)	1.171 ± 0.007 (s=403)	-0.3% .. +0.1%
Cartesian min max surface	2.310 ± 0.030 (s=193)	2.307 ± 0.019 (s=199)	-0.5% .. +0.2%
Spherical min max surface	7.230 ± 0.066 (s=65)	7.222 ± 0.031 (s=62)	-0.5% .. +0.3%

[gassmoeller]

Looks reasonable to me. However coveralls is failing, and the doc tester is failing with this warning:

/home/docs/checkouts/readthedocs.org/user_builds/gwb/checkouts/704/doc/sphinx/user_manual/cookbooks/index.md:8: WARNING: 'toctree': Invalid options format: expected key to start at column 0 [myst.directive_option]

[MFraters]

the coveralls stated a rate limiting, so I am just rerunning it and that seems to work for now. I will have to take a closer look at the warning though at some point. I am not sure why the doctester is failing though. I tried to fix it, but I didn't manage yet...

[gassmoeller]

I will let you judge which of these testers should be fixed before merging. I think the code change itself is correct.

[MFraters]

Since the readthedocs problem is unrelated to this pull request (I probably should have split the other fixes into separate pull requests to start with) and I agree that the code chagnes are correct, I will merge this and make a separate issue for the readthedocs compilation problem. Thanks for the review @gassmoeller!