Add fixed NNDVI

[Anmol-Srivastava]

Adds

• NNSpacePartitioner, NNDVI partitioning and detection algorithms
• examples, documentation, tests

Fixes

• ongoing false/indiscriminate alarm issue with NNDVI

Closes #36

[Anmol-Srivastava]

@tms-bananaquit I think you can begin review of this PR while I work out remaining issues with formatting/coverage. The major thing I anticipate you'll notice is the hard-coded unit tests for some of these compute functions (and elsewhere), for which I couldn't really find an alternative for now

[tms-bananaquit]

The sklearn dependency is already listed as part of install_requires, so we shouldn't need to add it to the other settings in setup.cfg. The logs show the build step is pulling python ver 3.10.8 instead of 3.10.7, and the sklearn .tar is a different file, but installation is apparently finishing successfully.

Apparently installing from the sklearn alias has been deprecated for a while, so I switched the dependency to scikit-learn. That fixed the build issue. Now there are some unit tests failing for other modules, which doesn't make a whole lot of sense given what code has(n't) been modified.

[Anmol-Srivastava]

The only full failure I see is the coverage 100% due to un-tested lines in nndvi.py -- addressing that now

[Anmol-Srivastava]

The mu == 0 case was removed because it's effectively no longer possible with the new change (and was cutting warnings in a very unlikely case to begin with). The remaining line to be covered is just the if drift, reset line which for some reason isn't being counted by one of the unit tests.

[Anmol-Srivastava]

Should be ready for review now

[Anmol-Srivastava]

Did the validation / docstring changes, waiting on some example business to resolve that comment too.

[Anmol-Srivastava]

Pushed the NNDVI-updated notebook for the time being, ~having some difficulty checking its look on my end~ the added code segment executes but it's taking a while to get output