Check for non-numeric keys in value_map in reclassify raster function #419

[claire-simpson]

Check that the value_map does not contain any non-numeric keys in the raster reclassify function. This function will fail if there are any None/NA keys in the value_map dictionary so instead we're explicitly checking for them upfront to provide the user with a clearer error message. Note that this still allows numpy.nan keys to pass. I've also created a test to confirm this check works as anticipated.

Fixes #419

[davemfish]

Thanks, @claire-simpson , this is very succinct and effective. Would you please add a note in HISTORY about the new TypeError being raised? And I found one little redundant line in the test.

[claire-simpson]

Thanks @davemfish! I noticed that many of the geoprocessing tests define the target_nodata two times, may be something to look into?

[davemfish]

Thanks @davemfish! I noticed that many of the geoprocessing tests define the target_nodata two times, may be something to look into?

Interesting, feel free to clean that up for us if you'd like to! It could happen in this PR if you want, or a separate one.

[claire-simpson]

Interesting, feel free to clean that up for us if you'd like to! It could happen in this PR if you want, or a separate one.

Went ahead and did it. I suspect target_nodata was redefined explicitly as the function called in between the duplicate assignments did set target_nodata but that should be out of scope and therefore unnecessary to reassign target_nodata in the test script

[davemfish]

Went ahead and did it. I suspect target_nodata was redefined explicitly as the function called in between the duplicate assignments did set target_nodata but that should be out of scope and therefore unnecessary to reassign target_nodata in the test script

Thank you!