BUG: fix indexing error in fill_region_float

[chrishavlin]

I was trying to figure out why the final row of pixels in the following plot seem to have elevated values:

import yt 
import matplotlib.pyplot as plt 
import numpy as np
ds = yt.load_sample("DeeplyNestedZoom")

le = ds.domain_left_edge
re = ds.arr([0.1,]*3, 'code_length')
res = (100,)*3
reg = ds.r[le[0]:re[0]:res[0]*1j, 
           le[1]:re[1]:res[1]*1j, 
           le[2]:re[2]:res[2]*1j,]
dens = reg['gas', 'density']
plt.imshow(np.log10(dens[:,50,:]), origin='lower')

The elevated values are not super obvious here... but i've been doing some experiments stitching together arbitrary grid objects into zarr files (over in yt-project/yt_experiments#10) and found some artifacts aligned with the boundaries of my chunks:

In any case, reading through fill_region_float, I noticed this line where there seems to be an indexing error. With the change on this branch, the above plots become

[chrishavlin]

wouldn't be surprised if an answer test or two fails...

[matthewturk]

wow. This is definitely a correction. Thank you.