Fix geotiff writer ignoring dtype argument

[BENR0]

This fixes #1730

• Closes geotiff writer ignores dtype argument, always writes float if enhance=False #1730
• Tests added

[codecov]

Codecov Report

Merging #1733 (26085e1) into main (7b8eb8d) will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##             main    #1733   +/-   ##
=======================================
  Coverage   92.83%   92.83%           
=======================================
  Files         260      260           
  Lines       38552    38561    +9     
=======================================
+ Hits        35791    35800    +9     
  Misses       2761     2761           

Flag	Coverage Δ
behaviourtests	4.81% <0.00%> (-0.01%)	⬇️
unittests	93.38% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
satpy/tests/writer_tests/test_geotiff.py	98.03% <100.00%> (+0.16%)	⬆️
satpy/writers/geotiff.py	91.07% <100.00%> (+0.16%)	⬆️

---

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 7b8eb8d...26085e1. Read the comment docs.

[coveralls]

Coverage increased (+0.002%) to 93.327% when pulling 26085e1 on BENR0:fix_geotiff_writer_dtype_bug into 7b8eb8d on pytroll:main.

[djhoese]

While this definitely is a bug it seems to me that saving an unenhanced image with "unit8" is in most cases not really wanted because there will be clipping of the data. Maybe a warning should be added for the case when clipping possibly might occur?

I'm inclined to avoid the "annoying" warning and assume that the user understands what they are saying when they do this. But I don't feel that strongly. I'm also not sure how common this "most cases" is. For reflectance data this wouldn't be that bad (major loss of precision, but still). For BT data, yeah not good. For RGBs, yeah not great either probably.

If you want to add the warning go for it. If @mraspaud or @pnuu or @gerritholl have feelings about this, speak now.

[BENR0]

I'm inclined to avoid the "annoying" warning and assume that the user understands what they are saying when they do this

I totally agree. I just suggested it because some users might not be aware and would wonder why their image looks so weird :-D. So if anyone want's me to add it I will, otherwise I am ok with how it is.

[gerritholl]

In DWD we've been generating images with stretch="no" but I don't know how we've prevented problems with values not between 0 and 256.

[djhoese]

Seems I lost track of this. Thanks again @BENR0.