Fix tiling issue in raster_tile.py

[srmsoumya]

Description

Raster tiler is missing a few tiles while creating the chips. Rounding the profile["transform"] to 3 decimal places is creating the issue of chips overwrite, changing this to 6 decimal places fixes the issue.

Type of change

• Bug fix (non-breaking change which fixes an issue)

Checklist:

• My PR has a descriptive title
• My code follows PEP8
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new errors
• I have added tests that prove my fix is effective or that my feature works
• My PR passes Travis CI tests
• My PR does not reduce coverage in Codecov

If your PR does not fulfill all of the requirements in the checklist above, that's OK! Just prepend [WIP] to the PR title until they are all satisfied. If you need help, @-mention a maintainer and/or add the Status: Help Needed label.

[rbavery]

Now that status checks are showing it's clear tests are not passing yet. This is strange, because they pass when I run them using make test locally. I think we need to align our gh action testing with our local testing. @srmsoumya @rodrigoalmeida94 I'm off for the day but if either of you get a chance can you investigate why tests are not passing on gh actions but they are with make test?

[srmsoumya]

@rbavery I was able to reproduce the test errors locally.

The array values are different for tiles in

• tests/data/rastertile_test_fill_nodata_expected/
• tests/data/rastertile_test_fill_nodata_result/

Visually they look similar but array values are different.
eg:

We are just changing the precision for file names, everything else remains the same.
Is it because we changed from

-  opencv-python>=4.1
+  opencv>=4.5.5

[rbavery]

@srmsoumya looking at your recent commit and just a heads up, using opencv appears to be what works for recent releases so I think we should stick to that. See #461