Inconsistent initialization of electric field-dependent quantities when E_field is initialized from j_tot

[hoppe93]

It turns out that when E_field is initialized from j_tot (i.e. a current density profile is prescribed), and f_re is initialized according to an avalanche distribution, the distribution function will end up initialized with E_field=0. The cause for this is the fact that to avoid a circular dependency, E_field must be initialized to 0 at the start of the initialization cycle, after which RunawayFluid can be initialized, followed again by an initialization of E_field from j_tot and the conductivity (contained in RunawayFluid).

Since E_field will be initialized at the start of the initialization, and since j_tot depends on j_re (and thereby f_re), f_re will be initialized with E_field=0 and therefore become zero.

A better approach to initializing f_re would be to allow the user to provide the E_field profile to use. Yet another alternative could be to separate the conductivity from RunawayFluid, as suggested in #84.