Investigate issues with capturing vertices' exposure at time of adoption with `node_threshold()`

[henriquesposito]

Although our node_threshold() function is more flexible (i.e. works on various types of networks), results are not consistent with those of netdiffuseR. We should investigate and fix or explain why this happens.

For example:

library(netdiffuseR) c(netdiffuseR::threshold(medInnovationsDiffNet))

vs.

library(manynet) unname(c(node_thresholds(as_diff_model(medInnovationsDiffNet))))

[jhollway]

A sweep of every exported function over fict_marvel turned up five more
diffusion failures, which look like they belong with the workover this family
needs rather than in a separate issue:

node_is_infected      'names' attribute [194] must be the same length as the vector
node_is_latent        'names' attribute [194] must be the same length as the vector
node_is_recovered     'names' attribute [194] must be the same length as the vector
node_in_adopter       no applicable method for 'filter' applied to an object of class 'stocnet'
node_x_exposure       no applicable method for 'as_diffusion' applied to an object of class 'stocnet'

The first three are the shape that CONTRIBUTING warns about: nodes may change
over waves while the ties do not, so a nodeset and a tieset of matching length
cannot be assumed. The last two report that no diffusion method applies to a
plain network, which is true but is not what a user reading the message would
take from it.

None of these is a signed or a multilevel problem, so they are not covered by
#25 or #26.

[jhollway]

A second sweep, over fict_actually, adds five network-level diffusion
functions with the same shape:

net_by_immunity              no applicable method for 'group_by' applied to an object of class 'stocnet'
net_by_infection_complete    no applicable method for 'group_by' applied to an object of class 'stocnet'
net_by_infection_peak        no applicable method for 'group_by' applied to an object of class 'stocnet'
net_by_recovery              no applicable method for 'group_by' applied to an object of class 'stocnet'
net_by_reproduction          no applicable method for 'group_by' applied to an object of class 'stocnet'

As with node_in_adopter() and node_x_exposure(), the network is simply not
a diffusion, so the right answer is probably a guard that says so and names
play_diffusion(), rather than a dplyr method reporting that it cannot
dispatch.