Improve SVM ability to use a contour-based shapefile as a source

[gwlucastrig]

For bathymetry data, contour based shapefiles consisting of line-based features (type Polyline and PolylineZ) often define both the shoreline and depth contours. Depth contours can be used directly as a source of depth soundings (an existing feature), but special handling is required to determine the boundary of lake. In cases where a bounding polygon shapefile is available, SVM should screen out the zero-depth (shoreline) contours from the line-based shapefile and use the lake-boundry features from the polygon shapefile. In cases where a bounding polygon shapefile is not available, it should use the zero-depth contour from the line-feature shapefile to define the lake boundaries (including any embedded island features).

One key requirement for using line-based features as bounding polygons is that the contours must form a closed loop (so that they can be readily promoted from line features to polygon features). In viewing some data sources, I've see examples where the contours would form a closed loop except that they are broken into multiple pieces. So this implementation must detect the pieces and stitch them together. It must also be able to detect "nested" contours (such as islands) and orient them correctly so that SVM can separate water regions from land regions. This situation arises in the Michigan inland-lake contours data product for several bodies of water, including Ford Lake.

[gwlucastrig]

I've completed the ability to use a contour based file as an input source. Code is submitted.

I am debating whether the ability to use contours as bounding polygons is actually useful. I've implemented a couple of test classes that do this, but it seems to me that in most cases a bounding polygon digitized from aerial photos or other survey techniques would (a) probably be available and (b) would be a better source of data.