feat: Support grayscale photometric interpretation

[kylebarron]

Closes #175

Change list

• Create new BlackIsZero and WhiteIsZero GPU shader modules
• Ensure that typed array type matches what the GPU expects for the given texture format. This had been preventing uint16 images from rendering

uint8_1band_deflate_block128_unaligned_mask.tif from developmentseed/geotiff-test-data#50

And with uint16 1-band data https://github.com/developmentseed/geotiff-test-data/blob/7fa07ac10f9d54b2ac2e557117432cf8f2f80eef/rasterio_generated/fixtures/uint16_1band_scale_offset_info.md#L4

[mdsumner]

The functions black_zero_to_rgb(float value) and white_zero_to_rgb(float value) expect a float, but color is a vec4.

[mdsumner]

that was tested in concert with PRs

#180 and #182

https://projects.pawsey.org.au/image-cogs/images/gebco_boxmerc_8192_UInt16.tif

created by

gdal_translate /vsicurl/https://projects.pawsey.org.au/idea-gebco-tif/GEBCO_2024.tif -scale -ot UInt16 -outsize 8192 4096 -of COG -co OVERVIEW_COUNT=3 /tmp/mycogs/gebco_8192_UInt16.tif

[kylebarron]

The functions black_zero_to_rgb(float value) and white_zero_to_rgb(float value) expect a float, but color is a vec4.

Yes, I might not have pushed my last commit on this branch