FIX: Coerce data types on writing GIFTI DataArrays

[effigies]

Starting with a test. I expect uniform failure.

cc @hyperbolicTom

Fixes #792.

[codecov]

Codecov Report

Merging #806 into maint/2.5.x will decrease coverage by 0.02%.
The diff coverage is 66.66%.

@@               Coverage Diff               @@
##           maint/2.5.x     #806      +/-   ##
===============================================
- Coverage           90%   89.98%   -0.03%     
===============================================
  Files               95       95              
  Lines            12048    12046       -2     
  Branches          2142     2143       +1     
===============================================
- Hits             10844    10839       -5     
- Misses             857      859       +2     
- Partials           347      348       +1

Impacted Files	Coverage Δ
nibabel/gifti/gifti.py	94.91% <66.66%> (-0.33%)	⬇️
nibabel/viewers.py	95.4% <0%> (-0.66%)	⬇️
nibabel/nicom/dicomwrappers.py	90.2% <0%> (-0.08%)	⬇️
nibabel/cmdline/parrec2nii.py	32.73% <0%> (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 f47f80e...2838c49. Read the comment docs.

[effigies]

Anybody feeling up for a review?