fix prob routing algorithm

[mpadge]

compare the current version which seems to give some odd results with the code of gdistance.

[mpadge]

@karpfen so i've managed to get the toy examples from the paper roughly working but:

1. Even they don't precisely reproduce the results given in the paper (although close enough to probably be okay), and
2. It doesn't work at all when scaled up to your street network example.
   I'll have to keep digging around ...

[karpfen]

Okay, cool! I didn't manage to break up the functions in a way that actually fixes any issues yet, but if you're feeling adventurous, here's a way that allows you to make a compact graph and pass it to the generic osm_router () function. Maybe that could be useful for you.

devtools::load_all (export_all = FALSE)
library (magrittr)
dat <- readRDS ("tests/compact-ways-munich.Rda") %>% makeCompactGraph %>%
    extract2 ("compact") %>% extract (c ("from_id", "to_id", "d_weighted"))

dat$from_id %<>% as.character %>% as.numeric
dat$to_id %<>% as.character %>% as.numeric
allids <- c (dat$from_id, dat$to_id) %>% sort %>% unique 
dat$from_id <- vapply (dat$from_id, function (x) { which (allids == x) -1 }, 0.)
dat$to_id <- vapply (dat$to_id, function (x) { which (allids == x) -1 }, 0.)

st <- dat$from_id [1]
en <- dat$to_id [30]
osm_router (dat, st, en)

[mpadge]

@karpfen check it out:

👍

[mpadge]

and then some using the full graph, first with no entropy

and then one with some randomness

😄 🎆 🍾 👍 💯 🍻 emoji fever!

[mpadge]

all done but sole remaining issue regarding the actual routing algorithm of #30