IP model segments and gradient calculations

[smnorris]

"segments" for IP models are currently arbitrary length source FWA geometries, further split at all features of interest. (streams need to be split at observations, barriers, etc, anything that we want to report on being upstream or downstream of a given stream in the model output).

Barring changes to mapping/landscape, slope of a given stream segment should obviously be constant over time rather than changing as model inputs change. Also, potential spawning/rearing habitat output is likely lost when gradient is averaged over longer distances often present in source geometries (for example, a stream rising from flat valley bottom up a mountainside that does not encounter any confluences or edge type changes). Instead, measure slope for IP models via one of these options:

1. using the stream profile directly (ie the stream profile generated by stream vertices)
2. break the stream into "gradient reaches" of similar slope, as per Cavalli/Tompalski of segmenting at local extremes of residuals of line of best fit
3. break the stream in to equidistant intervals and calculate average slope over the interval