Precise nearest neighbor mapping for extremely close points (e.g., weirs/levess)

[WPringle]

adding precise option (calls Matlab's native knnsearch) in nearest_neighbor_map and calling in the carry over weirs routine since sometimes the weir points can be so close together that the precision of ANN ourKNNsearch fails while matlab's knnsearch works

I added a check for the presence of knnsearch in case the user doesn't have it.

[krober10nd]

wow how thin are your weirs that this is a problem?

[WPringle]

So they made them on top of each other but to get it work with OM2D I moved them slightly by around ~1e-6 deg

[krober10nd]

ah okay 10 cm. seems reasonable :]

[krober10nd]

Did you see the small change in #206 ? I had to make these arrays column otherwise I could not carry over.

[WPringle]

Added code for toggling plotting of fort.24 in Make_f24 and fixing slope_calc bug in #208

[krober10nd]

Is this completed to warrant running all the tests?

[WPringle]

yes it's completed

[krober10nd]

Running Example_2 produces the following figure with several incorrectly labeled ocean boundaries.

[WPringle]

Running Example_2 produces the following figure with several incorrectly labeled ocean boundaries.

right but this is not related to this PR right

[WPringle]

I'll make a new PR for improving the make_bc

[krober10nd]

I guess. I thought that we had used OurKnnsearch for labeling boundaries.

[WPringle]

this PR doesn't use that change. Only for carrying over weirs. Otherwise the "approximate" nearest neighbor (i.e. ourKNNsearch) is used.

[krober10nd]

All tests/examples worked out (we'll need to do a different PR to highlight the API change in make_bc that we talked about on Slack). I could not test the global example on my laptop due to insufficient RAM.